diff --git a/meson.build b/meson.build index ca3ce8d..076f8f3 100644 --- a/meson.build +++ b/meson.build @@ -111,3 +111,15 @@ pkgconfig.generate( ] + platform_cflags, libraries: libwebrtc_audio_processing, ) + +pkgconfig.generate( + name: 'webrtc-audio-coding', + description: 'WebRTC Audio Coding library', + version: meson.project_version(), + filebase: 'webrtc-audio-coding', + subdirs: 'webrtc_audio_processing', + extra_cflags: [ + '-DWEBRTC_AUDIO_PROCESSING_ONLY_BUILD', + ] + platform_cflags, + libraries: libwebrtc_audio_coding, +) diff --git a/webrtc/modules/audio_coding/codecs/audio_decoder.cc b/webrtc/modules/audio_coding/codecs/audio_decoder.cc new file mode 100644 index 0000000..08d101c --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/audio_decoder.cc @@ -0,0 +1,106 @@ +/* + * Copyright (c) 2012 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#include "webrtc/modules/audio_coding/codecs/audio_decoder.h" + +#include + +#include "webrtc/base/checks.h" + +namespace webrtc { + +int AudioDecoder::Decode(const uint8_t* encoded, size_t encoded_len, + int sample_rate_hz, size_t max_decoded_bytes, + int16_t* decoded, SpeechType* speech_type) { + int duration = PacketDuration(encoded, encoded_len); + if (duration >= 0 && + duration * Channels() * sizeof(int16_t) > max_decoded_bytes) { + return -1; + } + return DecodeInternal(encoded, encoded_len, sample_rate_hz, decoded, + speech_type); +} + +int AudioDecoder::DecodeRedundant(const uint8_t* encoded, size_t encoded_len, + int sample_rate_hz, size_t max_decoded_bytes, + int16_t* decoded, SpeechType* speech_type) { + int duration = PacketDurationRedundant(encoded, encoded_len); + if (duration >= 0 && + duration * Channels() * sizeof(int16_t) > max_decoded_bytes) { + return -1; + } + return DecodeRedundantInternal(encoded, encoded_len, sample_rate_hz, decoded, + speech_type); +} + +int AudioDecoder::DecodeInternal(const uint8_t* encoded, size_t encoded_len, + int sample_rate_hz, int16_t* decoded, + SpeechType* speech_type) { + return kNotImplemented; +} + +int AudioDecoder::DecodeRedundantInternal(const uint8_t* encoded, + size_t encoded_len, + int sample_rate_hz, int16_t* decoded, + SpeechType* speech_type) { + return DecodeInternal(encoded, encoded_len, sample_rate_hz, decoded, + speech_type); +} + +bool AudioDecoder::HasDecodePlc() const { return false; } + +size_t AudioDecoder::DecodePlc(size_t num_frames, int16_t* decoded) { + return 0; +} + +int AudioDecoder::IncomingPacket(const uint8_t* payload, + size_t payload_len, + uint16_t rtp_sequence_number, + uint32_t rtp_timestamp, + uint32_t arrival_timestamp) { + return 0; +} + +int AudioDecoder::ErrorCode() { return 0; } + +int AudioDecoder::PacketDuration(const uint8_t* encoded, + size_t encoded_len) const { + return kNotImplemented; +} + +int AudioDecoder::PacketDurationRedundant(const uint8_t* encoded, + size_t encoded_len) const { + return kNotImplemented; +} + +bool AudioDecoder::PacketHasFec(const uint8_t* encoded, + size_t encoded_len) const { + return false; +} + +CNG_dec_inst* AudioDecoder::CngDecoderInstance() { + FATAL() << "Not a CNG decoder"; + return NULL; +} + +AudioDecoder::SpeechType AudioDecoder::ConvertSpeechType(int16_t type) { + switch (type) { + case 0: // TODO(hlundin): Both iSAC and Opus return 0 for speech. + case 1: + return kSpeech; + case 2: + return kComfortNoise; + default: + assert(false); + return kSpeech; + } +} + +} // namespace webrtc diff --git a/webrtc/modules/audio_coding/codecs/audio_decoder.h b/webrtc/modules/audio_coding/codecs/audio_decoder.h new file mode 100644 index 0000000..6189be0 --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/audio_decoder.h @@ -0,0 +1,123 @@ +/* + * Copyright (c) 2012 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#ifndef WEBRTC_MODULES_AUDIO_CODING_NETEQ_INCLUDE_AUDIO_DECODER_H_ +#define WEBRTC_MODULES_AUDIO_CODING_NETEQ_INCLUDE_AUDIO_DECODER_H_ + +#include // NULL + +#include "webrtc/base/constructormagic.h" +#include "webrtc/modules/audio_coding/codecs/cng/include/webrtc_cng.h" +#include "webrtc/typedefs.h" + +namespace webrtc { + +// This is the interface class for decoders in NetEQ. Each codec type will have +// and implementation of this class. +class AudioDecoder { + public: + enum SpeechType { + kSpeech = 1, + kComfortNoise = 2 + }; + + // Used by PacketDuration below. Save the value -1 for errors. + enum { kNotImplemented = -2 }; + + AudioDecoder() = default; + virtual ~AudioDecoder() = default; + + // Decodes |encode_len| bytes from |encoded| and writes the result in + // |decoded|. The maximum bytes allowed to be written into |decoded| is + // |max_decoded_bytes|. Returns the total number of samples across all + // channels. If the decoder produced comfort noise, |speech_type| + // is set to kComfortNoise, otherwise it is kSpeech. The desired output + // sample rate is provided in |sample_rate_hz|, which must be valid for the + // codec at hand. + virtual int Decode(const uint8_t* encoded, + size_t encoded_len, + int sample_rate_hz, + size_t max_decoded_bytes, + int16_t* decoded, + SpeechType* speech_type); + + // Same as Decode(), but interfaces to the decoders redundant decode function. + // The default implementation simply calls the regular Decode() method. + virtual int DecodeRedundant(const uint8_t* encoded, + size_t encoded_len, + int sample_rate_hz, + size_t max_decoded_bytes, + int16_t* decoded, + SpeechType* speech_type); + + // Indicates if the decoder implements the DecodePlc method. + virtual bool HasDecodePlc() const; + + // Calls the packet-loss concealment of the decoder to update the state after + // one or several lost packets. The caller has to make sure that the + // memory allocated in |decoded| should accommodate |num_frames| frames. + virtual size_t DecodePlc(size_t num_frames, int16_t* decoded); + + // Resets the decoder state (empty buffers etc.). + virtual void Reset() = 0; + + // Notifies the decoder of an incoming packet to NetEQ. + virtual int IncomingPacket(const uint8_t* payload, + size_t payload_len, + uint16_t rtp_sequence_number, + uint32_t rtp_timestamp, + uint32_t arrival_timestamp); + + // Returns the last error code from the decoder. + virtual int ErrorCode(); + + // Returns the duration in samples-per-channel of the payload in |encoded| + // which is |encoded_len| bytes long. Returns kNotImplemented if no duration + // estimate is available, or -1 in case of an error. + virtual int PacketDuration(const uint8_t* encoded, size_t encoded_len) const; + + // Returns the duration in samples-per-channel of the redandant payload in + // |encoded| which is |encoded_len| bytes long. Returns kNotImplemented if no + // duration estimate is available, or -1 in case of an error. + virtual int PacketDurationRedundant(const uint8_t* encoded, + size_t encoded_len) const; + + // Detects whether a packet has forward error correction. The packet is + // comprised of the samples in |encoded| which is |encoded_len| bytes long. + // Returns true if the packet has FEC and false otherwise. + virtual bool PacketHasFec(const uint8_t* encoded, size_t encoded_len) const; + + // If this is a CNG decoder, return the underlying CNG_dec_inst*. If this + // isn't a CNG decoder, don't call this method. + virtual CNG_dec_inst* CngDecoderInstance(); + + virtual size_t Channels() const = 0; + + protected: + static SpeechType ConvertSpeechType(int16_t type); + + virtual int DecodeInternal(const uint8_t* encoded, + size_t encoded_len, + int sample_rate_hz, + int16_t* decoded, + SpeechType* speech_type); + + virtual int DecodeRedundantInternal(const uint8_t* encoded, + size_t encoded_len, + int sample_rate_hz, + int16_t* decoded, + SpeechType* speech_type); + + private: + RTC_DISALLOW_COPY_AND_ASSIGN(AudioDecoder); +}; + +} // namespace webrtc +#endif // WEBRTC_MODULES_AUDIO_CODING_NETEQ_INCLUDE_AUDIO_DECODER_H_ diff --git a/webrtc/modules/audio_coding/codecs/audio_encoder.cc b/webrtc/modules/audio_coding/codecs/audio_encoder.cc new file mode 100644 index 0000000..6d76300 --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/audio_encoder.cc @@ -0,0 +1,55 @@ +/* + * Copyright (c) 2014 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#include "webrtc/modules/audio_coding/codecs/audio_encoder.h" +#include "webrtc/base/checks.h" + +namespace webrtc { + +AudioEncoder::EncodedInfo::EncodedInfo() = default; + +AudioEncoder::EncodedInfo::~EncodedInfo() = default; + +int AudioEncoder::RtpTimestampRateHz() const { + return SampleRateHz(); +} + +AudioEncoder::EncodedInfo AudioEncoder::Encode(uint32_t rtp_timestamp, + const int16_t* audio, + size_t num_samples_per_channel, + size_t max_encoded_bytes, + uint8_t* encoded) { + RTC_CHECK_EQ(num_samples_per_channel, + static_cast(SampleRateHz() / 100)); + EncodedInfo info = + EncodeInternal(rtp_timestamp, audio, max_encoded_bytes, encoded); + RTC_CHECK_LE(info.encoded_bytes, max_encoded_bytes); + return info; +} + +bool AudioEncoder::SetFec(bool enable) { + return !enable; +} + +bool AudioEncoder::SetDtx(bool enable) { + return !enable; +} + +bool AudioEncoder::SetApplication(Application application) { + return false; +} + +void AudioEncoder::SetMaxPlaybackRate(int frequency_hz) {} + +void AudioEncoder::SetProjectedPacketLossRate(double fraction) {} + +void AudioEncoder::SetTargetBitrate(int target_bps) {} + +} // namespace webrtc diff --git a/webrtc/modules/audio_coding/codecs/audio_encoder.h b/webrtc/modules/audio_coding/codecs/audio_encoder.h new file mode 100644 index 0000000..cda9d86 --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/audio_encoder.h @@ -0,0 +1,143 @@ +/* + * Copyright (c) 2014 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#ifndef WEBRTC_MODULES_AUDIO_CODING_CODECS_AUDIO_ENCODER_H_ +#define WEBRTC_MODULES_AUDIO_CODING_CODECS_AUDIO_ENCODER_H_ + +#include +#include + +#include "webrtc/typedefs.h" + +namespace webrtc { + +// This is the interface class for encoders in AudioCoding module. Each codec +// type must have an implementation of this class. +class AudioEncoder { + public: + struct EncodedInfoLeaf { + size_t encoded_bytes = 0; + uint32_t encoded_timestamp = 0; + int payload_type = 0; + bool send_even_if_empty = false; + bool speech = true; + }; + + // This is the main struct for auxiliary encoding information. Each encoded + // packet should be accompanied by one EncodedInfo struct, containing the + // total number of |encoded_bytes|, the |encoded_timestamp| and the + // |payload_type|. If the packet contains redundant encodings, the |redundant| + // vector will be populated with EncodedInfoLeaf structs. Each struct in the + // vector represents one encoding; the order of structs in the vector is the + // same as the order in which the actual payloads are written to the byte + // stream. When EncoderInfoLeaf structs are present in the vector, the main + // struct's |encoded_bytes| will be the sum of all the |encoded_bytes| in the + // vector. + struct EncodedInfo : public EncodedInfoLeaf { + EncodedInfo(); + ~EncodedInfo(); + + std::vector redundant; + }; + + virtual ~AudioEncoder() = default; + + // Returns the maximum number of bytes that can be produced by the encoder + // at each Encode() call. The caller can use the return value to determine + // the size of the buffer that needs to be allocated. This value is allowed + // to depend on encoder parameters like bitrate, frame size etc., so if + // any of these change, the caller of Encode() is responsible for checking + // that the buffer is large enough by calling MaxEncodedBytes() again. + virtual size_t MaxEncodedBytes() const = 0; + + // Returns the input sample rate in Hz and the number of input channels. + // These are constants set at instantiation time. + virtual int SampleRateHz() const = 0; + virtual int NumChannels() const = 0; + + // Returns the rate at which the RTP timestamps are updated. The default + // implementation returns SampleRateHz(). + virtual int RtpTimestampRateHz() const; + + // Returns the number of 10 ms frames the encoder will put in the next + // packet. This value may only change when Encode() outputs a packet; i.e., + // the encoder may vary the number of 10 ms frames from packet to packet, but + // it must decide the length of the next packet no later than when outputting + // the preceding packet. + virtual size_t Num10MsFramesInNextPacket() const = 0; + + // Returns the maximum value that can be returned by + // Num10MsFramesInNextPacket(). + virtual size_t Max10MsFramesInAPacket() const = 0; + + // Returns the current target bitrate in bits/s. The value -1 means that the + // codec adapts the target automatically, and a current target cannot be + // provided. + virtual int GetTargetBitrate() const = 0; + + // Accepts one 10 ms block of input audio (i.e., SampleRateHz() / 100 * + // NumChannels() samples). Multi-channel audio must be sample-interleaved. + // The encoder produces zero or more bytes of output in |encoded| and + // returns additional encoding information. + // The caller is responsible for making sure that |max_encoded_bytes| is + // not smaller than the number of bytes actually produced by the encoder. + // Encode() checks some preconditions, calls EncodeInternal() which does the + // actual work, and then checks some postconditions. + EncodedInfo Encode(uint32_t rtp_timestamp, + const int16_t* audio, + size_t num_samples_per_channel, + size_t max_encoded_bytes, + uint8_t* encoded); + + virtual EncodedInfo EncodeInternal(uint32_t rtp_timestamp, + const int16_t* audio, + size_t max_encoded_bytes, + uint8_t* encoded) = 0; + + // Resets the encoder to its starting state, discarding any input that has + // been fed to the encoder but not yet emitted in a packet. + virtual void Reset() = 0; + + // Enables or disables codec-internal FEC (forward error correction). Returns + // true if the codec was able to comply. The default implementation returns + // true when asked to disable FEC and false when asked to enable it (meaning + // that FEC isn't supported). + virtual bool SetFec(bool enable); + + // Enables or disables codec-internal VAD/DTX. Returns true if the codec was + // able to comply. The default implementation returns true when asked to + // disable DTX and false when asked to enable it (meaning that DTX isn't + // supported). + virtual bool SetDtx(bool enable); + + // Sets the application mode. Returns true if the codec was able to comply. + // The default implementation just returns false. + enum class Application { kSpeech, kAudio }; + virtual bool SetApplication(Application application); + + // Tells the encoder about the highest sample rate the decoder is expected to + // use when decoding the bitstream. The encoder would typically use this + // information to adjust the quality of the encoding. The default + // implementation just returns true. + virtual void SetMaxPlaybackRate(int frequency_hz); + + // Tells the encoder what the projected packet loss rate is. The rate is in + // the range [0.0, 1.0]. The encoder would typically use this information to + // adjust channel coding efforts, such as FEC. The default implementation + // does nothing. + virtual void SetProjectedPacketLossRate(double fraction); + + // Tells the encoder what average bitrate we'd like it to produce. The + // encoder is free to adjust or disregard the given bitrate (the default + // implementation does the latter). + virtual void SetTargetBitrate(int target_bps); +}; +} // namespace webrtc +#endif // WEBRTC_MODULES_AUDIO_CODING_CODECS_AUDIO_ENCODER_H_ diff --git a/webrtc/modules/audio_coding/codecs/cng/include/webrtc_cng.h b/webrtc/modules/audio_coding/codecs/cng/include/webrtc_cng.h new file mode 100644 index 0000000..35660c4 --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/cng/include/webrtc_cng.h @@ -0,0 +1,163 @@ +/* + * Copyright (c) 2012 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + + +#ifndef WEBRTC_MODULES_AUDIO_CODING_CODECS_CNG_MAIN_INCLUDE_WEBRTC_CNG_H_ +#define WEBRTC_MODULES_AUDIO_CODING_CODECS_CNG_MAIN_INCLUDE_WEBRTC_CNG_H_ + +#include +#include "webrtc/typedefs.h" + +#ifdef __cplusplus +extern "C" { +#endif + +#define WEBRTC_CNG_MAX_LPC_ORDER 12 +#define WEBRTC_CNG_MAX_OUTSIZE_ORDER 640 + +/* Define Error codes. */ + +/* 6100 Encoder */ +#define CNG_ENCODER_NOT_INITIATED 6120 +#define CNG_DISALLOWED_LPC_ORDER 6130 +#define CNG_DISALLOWED_FRAME_SIZE 6140 +#define CNG_DISALLOWED_SAMPLING_FREQUENCY 6150 +/* 6200 Decoder */ +#define CNG_DECODER_NOT_INITIATED 6220 + +typedef struct WebRtcCngEncInst CNG_enc_inst; +typedef struct WebRtcCngDecInst CNG_dec_inst; + +/**************************************************************************** + * WebRtcCng_CreateEnc/Dec(...) + * + * These functions create an instance to the specified structure + * + * Input: + * - XXX_inst : Pointer to created instance that should be created + * + * Return value : 0 - Ok + * -1 - Error + */ +int16_t WebRtcCng_CreateEnc(CNG_enc_inst** cng_inst); +int16_t WebRtcCng_CreateDec(CNG_dec_inst** cng_inst); + +/**************************************************************************** + * WebRtcCng_InitEnc/Dec(...) + * + * This function initializes a instance + * + * Input: + * - cng_inst : Instance that should be initialized + * + * - fs : 8000 for narrowband and 16000 for wideband + * - interval : generate SID data every interval ms + * - quality : Number of refl. coefs, maximum allowed is 12 + * + * Output: + * - cng_inst : Initialized instance + * + * Return value : 0 - Ok + * -1 - Error + */ + +int WebRtcCng_InitEnc(CNG_enc_inst* cng_inst, int fs, int16_t interval, + int16_t quality); +void WebRtcCng_InitDec(CNG_dec_inst* cng_inst); + +/**************************************************************************** + * WebRtcCng_FreeEnc/Dec(...) + * + * These functions frees the dynamic memory of a specified instance + * + * Input: + * - cng_inst : Pointer to created instance that should be freed + * + * Return value : 0 - Ok + * -1 - Error + */ +int16_t WebRtcCng_FreeEnc(CNG_enc_inst* cng_inst); +int16_t WebRtcCng_FreeDec(CNG_dec_inst* cng_inst); + +/**************************************************************************** + * WebRtcCng_Encode(...) + * + * These functions analyzes background noise + * + * Input: + * - cng_inst : Pointer to created instance + * - speech : Signal to be analyzed + * - nrOfSamples : Size of speech vector + * - forceSID : not zero to force SID frame and reset + * + * Output: + * - bytesOut : Nr of bytes to transmit, might be 0 + * + * Return value : 0 - Ok + * -1 - Error + */ +int WebRtcCng_Encode(CNG_enc_inst* cng_inst, int16_t* speech, + size_t nrOfSamples, uint8_t* SIDdata, + size_t* bytesOut, int16_t forceSID); + +/**************************************************************************** + * WebRtcCng_UpdateSid(...) + * + * These functions updates the CN state, when a new SID packet arrives + * + * Input: + * - cng_inst : Pointer to created instance that should be freed + * - SID : SID packet, all headers removed + * - length : Length in bytes of SID packet + * + * Return value : 0 - Ok + * -1 - Error + */ +int16_t WebRtcCng_UpdateSid(CNG_dec_inst* cng_inst, uint8_t* SID, + size_t length); + +/**************************************************************************** + * WebRtcCng_Generate(...) + * + * These functions generates CN data when needed + * + * Input: + * - cng_inst : Pointer to created instance that should be freed + * - outData : pointer to area to write CN data + * - nrOfSamples : How much data to generate + * - new_period : >0 if a new period of CNG, will reset history + * + * Return value : 0 - Ok + * -1 - Error + */ +int16_t WebRtcCng_Generate(CNG_dec_inst* cng_inst, int16_t* outData, + size_t nrOfSamples, int16_t new_period); + +/***************************************************************************** + * WebRtcCng_GetErrorCodeEnc/Dec(...) + * + * This functions can be used to check the error code of a CNG instance. When + * a function returns -1 a error code will be set for that instance. The + * function below extract the code of the last error that occurred in the + * specified instance. + * + * Input: + * - CNG_inst : CNG enc/dec instance + * + * Return value : Error code + */ +int16_t WebRtcCng_GetErrorCodeEnc(CNG_enc_inst* cng_inst); +int16_t WebRtcCng_GetErrorCodeDec(CNG_dec_inst* cng_inst); + +#ifdef __cplusplus +} +#endif + +#endif // WEBRTC_MODULES_AUDIO_CODING_CODECS_CNG_MAIN_INCLUDE_WEBRTC_CNG_H_ diff --git a/webrtc/modules/audio_coding/codecs/isac/audio_decoder_isac_t.h b/webrtc/modules/audio_coding/codecs/isac/audio_decoder_isac_t.h new file mode 100644 index 0000000..845af42 --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/audio_decoder_isac_t.h @@ -0,0 +1,54 @@ +/* + * Copyright (c) 2015 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#ifndef WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_AUDIO_DECODER_ISAC_T_H_ +#define WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_AUDIO_DECODER_ISAC_T_H_ + +#include + +#include "webrtc/modules/audio_coding/codecs/audio_decoder.h" +#include "webrtc/modules/audio_coding/codecs/isac/locked_bandwidth_info.h" + +namespace webrtc { + +template +class AudioDecoderIsacT final : public AudioDecoder { + public: + AudioDecoderIsacT(); + explicit AudioDecoderIsacT(LockedIsacBandwidthInfo* bwinfo); + ~AudioDecoderIsacT() override; + + bool HasDecodePlc() const override; + size_t DecodePlc(size_t num_frames, int16_t* decoded) override; + void Reset() override; + int IncomingPacket(const uint8_t* payload, + size_t payload_len, + uint16_t rtp_sequence_number, + uint32_t rtp_timestamp, + uint32_t arrival_timestamp) override; + int ErrorCode() override; + size_t Channels() const override; + int DecodeInternal(const uint8_t* encoded, + size_t encoded_len, + int sample_rate_hz, + int16_t* decoded, + SpeechType* speech_type) override; + + private: + typename T::instance_type* isac_state_; + LockedIsacBandwidthInfo* bwinfo_; + int decoder_sample_rate_hz_; + + RTC_DISALLOW_COPY_AND_ASSIGN(AudioDecoderIsacT); +}; + +} // namespace webrtc + +#endif // WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_AUDIO_DECODER_ISAC_T_H_ diff --git a/webrtc/modules/audio_coding/codecs/isac/audio_decoder_isac_t_impl.h b/webrtc/modules/audio_coding/codecs/isac/audio_decoder_isac_t_impl.h new file mode 100644 index 0000000..a986bc4 --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/audio_decoder_isac_t_impl.h @@ -0,0 +1,104 @@ +/* + * Copyright (c) 2015 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#ifndef WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_AUDIO_DECODER_ISAC_T_IMPL_H_ +#define WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_AUDIO_DECODER_ISAC_T_IMPL_H_ + +#include "webrtc/modules/audio_coding/codecs/isac/main/include/audio_decoder_isac.h" + +#include "webrtc/base/checks.h" + +namespace webrtc { + +template +AudioDecoderIsacT::AudioDecoderIsacT() + : AudioDecoderIsacT(nullptr) {} + +template +AudioDecoderIsacT::AudioDecoderIsacT(LockedIsacBandwidthInfo* bwinfo) + : bwinfo_(bwinfo), decoder_sample_rate_hz_(-1) { + RTC_CHECK_EQ(0, T::Create(&isac_state_)); + T::DecoderInit(isac_state_); + if (bwinfo_) { + IsacBandwidthInfo bi; + T::GetBandwidthInfo(isac_state_, &bi); + bwinfo_->Set(bi); + } +} + +template +AudioDecoderIsacT::~AudioDecoderIsacT() { + RTC_CHECK_EQ(0, T::Free(isac_state_)); +} + +template +int AudioDecoderIsacT::DecodeInternal(const uint8_t* encoded, + size_t encoded_len, + int sample_rate_hz, + int16_t* decoded, + SpeechType* speech_type) { + RTC_CHECK(sample_rate_hz == 16000 || sample_rate_hz == 32000) + << "Unsupported sample rate " << sample_rate_hz; + if (sample_rate_hz != decoder_sample_rate_hz_) { + RTC_CHECK_EQ(0, T::SetDecSampRate(isac_state_, sample_rate_hz)); + decoder_sample_rate_hz_ = sample_rate_hz; + } + int16_t temp_type = 1; // Default is speech. + int ret = + T::DecodeInternal(isac_state_, encoded, encoded_len, decoded, &temp_type); + *speech_type = ConvertSpeechType(temp_type); + return ret; +} + +template +bool AudioDecoderIsacT::HasDecodePlc() const { + return false; +} + +template +size_t AudioDecoderIsacT::DecodePlc(size_t num_frames, int16_t* decoded) { + return T::DecodePlc(isac_state_, decoded, num_frames); +} + +template +void AudioDecoderIsacT::Reset() { + T::DecoderInit(isac_state_); +} + +template +int AudioDecoderIsacT::IncomingPacket(const uint8_t* payload, + size_t payload_len, + uint16_t rtp_sequence_number, + uint32_t rtp_timestamp, + uint32_t arrival_timestamp) { + int ret = T::UpdateBwEstimate(isac_state_, payload, payload_len, + rtp_sequence_number, rtp_timestamp, + arrival_timestamp); + if (bwinfo_) { + IsacBandwidthInfo bwinfo; + T::GetBandwidthInfo(isac_state_, &bwinfo); + bwinfo_->Set(bwinfo); + } + return ret; +} + +template +int AudioDecoderIsacT::ErrorCode() { + return T::GetErrorCode(isac_state_); +} + +template +size_t AudioDecoderIsacT::Channels() const { + return 1; +} + +} // namespace webrtc + +#endif // WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_AUDIO_DECODER_ISAC_T_IMPL_H_ diff --git a/webrtc/modules/audio_coding/codecs/isac/audio_encoder_isac_t.h b/webrtc/modules/audio_coding/codecs/isac/audio_encoder_isac_t.h new file mode 100644 index 0000000..b15ad94 --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/audio_encoder_isac_t.h @@ -0,0 +1,97 @@ +/* + * Copyright (c) 2014 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#ifndef WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_AUDIO_ENCODER_ISAC_T_H_ +#define WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_AUDIO_ENCODER_ISAC_T_H_ + +#include + +#include "webrtc/modules/audio_coding/codecs/audio_encoder.h" +#include "webrtc/modules/audio_coding/codecs/isac/locked_bandwidth_info.h" + +namespace webrtc { + +struct CodecInst; + +template +class AudioEncoderIsacT final : public AudioEncoder { + public: + // Allowed combinations of sample rate, frame size, and bit rate are + // - 16000 Hz, 30 ms, 10000-32000 bps + // - 16000 Hz, 60 ms, 10000-32000 bps + // - 32000 Hz, 30 ms, 10000-56000 bps (if T has super-wideband support) + struct Config { + bool IsOk() const; + + LockedIsacBandwidthInfo* bwinfo = nullptr; + + int payload_type = 103; + int sample_rate_hz = 16000; + int frame_size_ms = 30; + int bit_rate = kDefaultBitRate; // Limit on the short-term average bit + // rate, in bits/s. + int max_payload_size_bytes = -1; + int max_bit_rate = -1; + + // If true, the encoder will dynamically adjust frame size and bit rate; + // the configured values are then merely the starting point. + bool adaptive_mode = false; + + // In adaptive mode, prevent adaptive changes to the frame size. (Not used + // in nonadaptive mode.) + bool enforce_frame_size = false; + }; + + explicit AudioEncoderIsacT(const Config& config); + explicit AudioEncoderIsacT(const CodecInst& codec_inst, + LockedIsacBandwidthInfo* bwinfo); + ~AudioEncoderIsacT() override; + + size_t MaxEncodedBytes() const override; + int SampleRateHz() const override; + int NumChannels() const override; + size_t Num10MsFramesInNextPacket() const override; + size_t Max10MsFramesInAPacket() const override; + int GetTargetBitrate() const override; + EncodedInfo EncodeInternal(uint32_t rtp_timestamp, + const int16_t* audio, + size_t max_encoded_bytes, + uint8_t* encoded) override; + void Reset() override; + + private: + // This value is taken from STREAM_SIZE_MAX_60 for iSAC float (60 ms) and + // STREAM_MAXW16_60MS for iSAC fix (60 ms). + static const size_t kSufficientEncodeBufferSizeBytes = 400; + + static const int kDefaultBitRate = 32000; + + // Recreate the iSAC encoder instance with the given settings, and save them. + void RecreateEncoderInstance(const Config& config); + + Config config_; + typename T::instance_type* isac_state_ = nullptr; + LockedIsacBandwidthInfo* bwinfo_ = nullptr; + + // Have we accepted input but not yet emitted it in a packet? + bool packet_in_progress_ = false; + + // Timestamp of the first input of the currently in-progress packet. + uint32_t packet_timestamp_; + + // Timestamp of the previously encoded packet. + uint32_t last_encoded_timestamp_; + + RTC_DISALLOW_COPY_AND_ASSIGN(AudioEncoderIsacT); +}; + +} // namespace webrtc + +#endif // WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_AUDIO_ENCODER_ISAC_T_H_ diff --git a/webrtc/modules/audio_coding/codecs/isac/audio_encoder_isac_t_impl.h b/webrtc/modules/audio_coding/codecs/isac/audio_encoder_isac_t_impl.h new file mode 100644 index 0000000..279f80d --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/audio_encoder_isac_t_impl.h @@ -0,0 +1,190 @@ +/* + * Copyright (c) 2014 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#ifndef WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_AUDIO_ENCODER_ISAC_T_IMPL_H_ +#define WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_AUDIO_ENCODER_ISAC_T_IMPL_H_ + +#include "webrtc/modules/audio_coding/codecs/isac/main/include/audio_encoder_isac.h" + +#include "webrtc/base/checks.h" + +namespace webrtc { + +template +typename AudioEncoderIsacT::Config CreateIsacConfig( + const CodecInst& codec_inst, + LockedIsacBandwidthInfo* bwinfo) { + typename AudioEncoderIsacT::Config config; + config.bwinfo = bwinfo; + config.payload_type = codec_inst.pltype; + config.sample_rate_hz = codec_inst.plfreq; + config.frame_size_ms = + rtc::CheckedDivExact(1000 * codec_inst.pacsize, config.sample_rate_hz); + config.adaptive_mode = (codec_inst.rate == -1); + if (codec_inst.rate != -1) + config.bit_rate = codec_inst.rate; + return config; +} + +template +bool AudioEncoderIsacT::Config::IsOk() const { + if (max_bit_rate < 32000 && max_bit_rate != -1) + return false; + if (max_payload_size_bytes < 120 && max_payload_size_bytes != -1) + return false; + if (adaptive_mode && !bwinfo) + return false; + switch (sample_rate_hz) { + case 16000: + if (max_bit_rate > 53400) + return false; + if (max_payload_size_bytes > 400) + return false; + return (frame_size_ms == 30 || frame_size_ms == 60) && + (bit_rate == 0 || (bit_rate >= 10000 && bit_rate <= 32000)); + case 32000: + if (max_bit_rate > 160000) + return false; + if (max_payload_size_bytes > 600) + return false; + return T::has_swb && + (frame_size_ms == 30 && + (bit_rate == 0 || (bit_rate >= 10000 && bit_rate <= 56000))); + default: + return false; + } +} + +template +AudioEncoderIsacT::AudioEncoderIsacT(const Config& config) { + RecreateEncoderInstance(config); +} + +template +AudioEncoderIsacT::AudioEncoderIsacT(const CodecInst& codec_inst, + LockedIsacBandwidthInfo* bwinfo) + : AudioEncoderIsacT(CreateIsacConfig(codec_inst, bwinfo)) {} + +template +AudioEncoderIsacT::~AudioEncoderIsacT() { + RTC_CHECK_EQ(0, T::Free(isac_state_)); +} + +template +size_t AudioEncoderIsacT::MaxEncodedBytes() const { + return kSufficientEncodeBufferSizeBytes; +} + +template +int AudioEncoderIsacT::SampleRateHz() const { + return T::EncSampRate(isac_state_); +} + +template +int AudioEncoderIsacT::NumChannels() const { + return 1; +} + +template +size_t AudioEncoderIsacT::Num10MsFramesInNextPacket() const { + const int samples_in_next_packet = T::GetNewFrameLen(isac_state_); + return static_cast( + rtc::CheckedDivExact(samples_in_next_packet, + rtc::CheckedDivExact(SampleRateHz(), 100))); +} + +template +size_t AudioEncoderIsacT::Max10MsFramesInAPacket() const { + return 6; // iSAC puts at most 60 ms in a packet. +} + +template +int AudioEncoderIsacT::GetTargetBitrate() const { + if (config_.adaptive_mode) + return -1; + return config_.bit_rate == 0 ? kDefaultBitRate : config_.bit_rate; +} + +template +AudioEncoder::EncodedInfo AudioEncoderIsacT::EncodeInternal( + uint32_t rtp_timestamp, + const int16_t* audio, + size_t max_encoded_bytes, + uint8_t* encoded) { + if (!packet_in_progress_) { + // Starting a new packet; remember the timestamp for later. + packet_in_progress_ = true; + packet_timestamp_ = rtp_timestamp; + } + if (bwinfo_) { + IsacBandwidthInfo bwinfo = bwinfo_->Get(); + T::SetBandwidthInfo(isac_state_, &bwinfo); + } + int r = T::Encode(isac_state_, audio, encoded); + RTC_CHECK_GE(r, 0) << "Encode failed (error code " + << T::GetErrorCode(isac_state_) << ")"; + + // T::Encode doesn't allow us to tell it the size of the output + // buffer. All we can do is check for an overrun after the fact. + RTC_CHECK_LE(static_cast(r), max_encoded_bytes); + + if (r == 0) + return EncodedInfo(); + + // Got enough input to produce a packet. Return the saved timestamp from + // the first chunk of input that went into the packet. + packet_in_progress_ = false; + EncodedInfo info; + info.encoded_bytes = r; + info.encoded_timestamp = packet_timestamp_; + info.payload_type = config_.payload_type; + return info; +} + +template +void AudioEncoderIsacT::Reset() { + RecreateEncoderInstance(config_); +} + +template +void AudioEncoderIsacT::RecreateEncoderInstance(const Config& config) { + RTC_CHECK(config.IsOk()); + packet_in_progress_ = false; + bwinfo_ = config.bwinfo; + if (isac_state_) + RTC_CHECK_EQ(0, T::Free(isac_state_)); + RTC_CHECK_EQ(0, T::Create(&isac_state_)); + RTC_CHECK_EQ(0, T::EncoderInit(isac_state_, config.adaptive_mode ? 0 : 1)); + RTC_CHECK_EQ(0, T::SetEncSampRate(isac_state_, config.sample_rate_hz)); + const int bit_rate = config.bit_rate == 0 ? kDefaultBitRate : config.bit_rate; + if (config.adaptive_mode) { + RTC_CHECK_EQ(0, T::ControlBwe(isac_state_, bit_rate, config.frame_size_ms, + config.enforce_frame_size)); + } else { + RTC_CHECK_EQ(0, T::Control(isac_state_, bit_rate, config.frame_size_ms)); + } + if (config.max_payload_size_bytes != -1) + RTC_CHECK_EQ( + 0, T::SetMaxPayloadSize(isac_state_, config.max_payload_size_bytes)); + if (config.max_bit_rate != -1) + RTC_CHECK_EQ(0, T::SetMaxRate(isac_state_, config.max_bit_rate)); + + // Set the decoder sample rate even though we just use the encoder. This + // doesn't appear to be necessary to produce a valid encoding, but without it + // we get an encoding that isn't bit-for-bit identical with what a combined + // encoder+decoder object produces. + RTC_CHECK_EQ(0, T::SetDecSampRate(isac_state_, config.sample_rate_hz)); + + config_ = config; +} + +} // namespace webrtc + +#endif // WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_AUDIO_ENCODER_ISAC_T_IMPL_H_ diff --git a/webrtc/modules/audio_coding/codecs/isac/locked_bandwidth_info.h b/webrtc/modules/audio_coding/codecs/isac/locked_bandwidth_info.h new file mode 100644 index 0000000..bbb040d --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/locked_bandwidth_info.h @@ -0,0 +1,45 @@ +/* + * Copyright (c) 2015 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#ifndef WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_LOCKED_BANDWIDTH_INFO_H_ +#define WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_LOCKED_BANDWIDTH_INFO_H_ + +#include "webrtc/base/scoped_ptr.h" +#include "webrtc/base/thread_annotations.h" +#include "webrtc/modules/audio_coding/codecs/isac/bandwidth_info.h" +#include "webrtc/system_wrappers/include/critical_section_wrapper.h" + +namespace webrtc { + +// An IsacBandwidthInfo that's safe to access from multiple threads because +// it's protected by a mutex. +class LockedIsacBandwidthInfo final { + public: + LockedIsacBandwidthInfo(); + ~LockedIsacBandwidthInfo(); + + IsacBandwidthInfo Get() const { + CriticalSectionScoped cs(lock_.get()); + return bwinfo_; + } + + void Set(const IsacBandwidthInfo& bwinfo) { + CriticalSectionScoped cs(lock_.get()); + bwinfo_ = bwinfo; + } + + private: + const rtc::scoped_ptr lock_; + IsacBandwidthInfo bwinfo_ GUARDED_BY(lock_); +}; + +} // namespace webrtc + +#endif // WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_LOCKED_BANDWIDTH_INFO_H_ diff --git a/webrtc/modules/audio_coding/codecs/isac/main/include/audio_decoder_isac.h b/webrtc/modules/audio_coding/codecs/isac/main/include/audio_decoder_isac.h new file mode 100644 index 0000000..dcd4852 --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/main/include/audio_decoder_isac.h @@ -0,0 +1,22 @@ +/* + * Copyright (c) 2015 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#ifndef WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_INCLUDE_AUDIO_DECODER_ISAC_H_ +#define WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_INCLUDE_AUDIO_DECODER_ISAC_H_ + +#include "webrtc/modules/audio_coding/codecs/isac/audio_decoder_isac_t.h" +#include "webrtc/modules/audio_coding/codecs/isac/main/source/isac_float_type.h" + +namespace webrtc { + +using AudioDecoderIsac = AudioDecoderIsacT; + +} // namespace webrtc +#endif // WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_INCLUDE_AUDIO_ENCODER_ISAC_H_ diff --git a/webrtc/modules/audio_coding/codecs/isac/main/include/audio_encoder_isac.h b/webrtc/modules/audio_coding/codecs/isac/main/include/audio_encoder_isac.h new file mode 100644 index 0000000..cc8665d --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/main/include/audio_encoder_isac.h @@ -0,0 +1,22 @@ +/* + * Copyright (c) 2014 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#ifndef WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_INCLUDE_AUDIO_ENCODER_ISAC_H_ +#define WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_INCLUDE_AUDIO_ENCODER_ISAC_H_ + +#include "webrtc/modules/audio_coding/codecs/isac/audio_encoder_isac_t.h" +#include "webrtc/modules/audio_coding/codecs/isac/main/source/isac_float_type.h" + +namespace webrtc { + +using AudioEncoderIsac = AudioEncoderIsacT; + +} // namespace webrtc +#endif // WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_INCLUDE_AUDIO_ENCODER_ISAC_H_ diff --git a/webrtc/modules/audio_coding/codecs/isac/main/source/audio_decoder_isac.cc b/webrtc/modules/audio_coding/codecs/isac/main/source/audio_decoder_isac.cc new file mode 100644 index 0000000..8e0603e --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/main/source/audio_decoder_isac.cc @@ -0,0 +1,20 @@ +/* + * Copyright (c) 2015 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#include "webrtc/modules/audio_coding/codecs/isac/main/include/audio_decoder_isac.h" + +#include "webrtc/modules/audio_coding/codecs/isac/audio_decoder_isac_t_impl.h" + +namespace webrtc { + +// Explicit instantiation: +template class AudioDecoderIsacT; + +} // namespace webrtc diff --git a/webrtc/modules/audio_coding/codecs/isac/main/source/audio_encoder_isac.cc b/webrtc/modules/audio_coding/codecs/isac/main/source/audio_encoder_isac.cc new file mode 100644 index 0000000..64b9815 --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/main/source/audio_encoder_isac.cc @@ -0,0 +1,20 @@ +/* + * Copyright (c) 2014 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#include "webrtc/modules/audio_coding/codecs/isac/main/include/audio_encoder_isac.h" + +#include "webrtc/modules/audio_coding/codecs/isac/audio_encoder_isac_t_impl.h" + +namespace webrtc { + +// Explicit instantiation: +template class AudioEncoderIsacT; + +} // namespace webrtc diff --git a/webrtc/modules/audio_coding/codecs/isac/main/source/bandwidth_estimator.c b/webrtc/modules/audio_coding/codecs/isac/main/source/bandwidth_estimator.c new file mode 100644 index 0000000..82fd053 --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/main/source/bandwidth_estimator.c @@ -0,0 +1,1031 @@ +/* + * Copyright (c) 2012 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +/* + * BwEstimator.c + * + * This file contains the code for the Bandwidth Estimator designed + * for iSAC. + * + */ + +#include "bandwidth_estimator.h" +#include "settings.h" +#include "webrtc/modules/audio_coding/codecs/isac/main/include/isac.h" + +#include +#include +#include + +/* array of quantization levels for bottle neck info; Matlab code: */ +/* sprintf('%4.1ff, ', logspace(log10(5000), log10(40000), 12)) */ +static const float kQRateTableWb[12] = +{ + 10000.0f, 11115.3f, 12355.1f, 13733.1f, 15264.8f, 16967.3f, + 18859.8f, 20963.3f, 23301.4f, 25900.3f, 28789.0f, 32000.0f}; + + +static const float kQRateTableSwb[24] = +{ + 10000.0f, 11115.3f, 12355.1f, 13733.1f, 15264.8f, 16967.3f, + 18859.8f, 20963.3f, 23153.1f, 25342.9f, 27532.7f, 29722.5f, + 31912.3f, 34102.1f, 36291.9f, 38481.7f, 40671.4f, 42861.2f, + 45051.0f, 47240.8f, 49430.6f, 51620.4f, 53810.2f, 56000.0f, +}; + + + + +int32_t WebRtcIsac_InitBandwidthEstimator( + BwEstimatorstr* bwest_str, + enum IsacSamplingRate encoderSampRate, + enum IsacSamplingRate decoderSampRate) +{ + switch(encoderSampRate) + { + case kIsacWideband: + { + bwest_str->send_bw_avg = INIT_BN_EST_WB; + break; + } + case kIsacSuperWideband: + { + bwest_str->send_bw_avg = INIT_BN_EST_SWB; + break; + } + } + + switch(decoderSampRate) + { + case kIsacWideband: + { + bwest_str->prev_frame_length = INIT_FRAME_LEN_WB; + bwest_str->rec_bw_inv = 1.0f / + (INIT_BN_EST_WB + INIT_HDR_RATE_WB); + bwest_str->rec_bw = (int32_t)INIT_BN_EST_WB; + bwest_str->rec_bw_avg_Q = INIT_BN_EST_WB; + bwest_str->rec_bw_avg = INIT_BN_EST_WB + INIT_HDR_RATE_WB; + bwest_str->rec_header_rate = INIT_HDR_RATE_WB; + break; + } + case kIsacSuperWideband: + { + bwest_str->prev_frame_length = INIT_FRAME_LEN_SWB; + bwest_str->rec_bw_inv = 1.0f / + (INIT_BN_EST_SWB + INIT_HDR_RATE_SWB); + bwest_str->rec_bw = (int32_t)INIT_BN_EST_SWB; + bwest_str->rec_bw_avg_Q = INIT_BN_EST_SWB; + bwest_str->rec_bw_avg = INIT_BN_EST_SWB + INIT_HDR_RATE_SWB; + bwest_str->rec_header_rate = INIT_HDR_RATE_SWB; + break; + } + } + + bwest_str->prev_rec_rtp_number = 0; + bwest_str->prev_rec_arr_ts = 0; + bwest_str->prev_rec_send_ts = 0; + bwest_str->prev_rec_rtp_rate = 1.0f; + bwest_str->last_update_ts = 0; + bwest_str->last_reduction_ts = 0; + bwest_str->count_tot_updates_rec = -9; + bwest_str->rec_jitter = 10.0f; + bwest_str->rec_jitter_short_term = 0.0f; + bwest_str->rec_jitter_short_term_abs = 5.0f; + bwest_str->rec_max_delay = 10.0f; + bwest_str->rec_max_delay_avg_Q = 10.0f; + bwest_str->num_pkts_rec = 0; + + bwest_str->send_max_delay_avg = 10.0f; + + bwest_str->hsn_detect_rec = 0; + + bwest_str->num_consec_rec_pkts_over_30k = 0; + + bwest_str->hsn_detect_snd = 0; + + bwest_str->num_consec_snt_pkts_over_30k = 0; + + bwest_str->in_wait_period = 0; + + bwest_str->change_to_WB = 0; + + bwest_str->numConsecLatePkts = 0; + bwest_str->consecLatency = 0; + bwest_str->inWaitLatePkts = 0; + bwest_str->senderTimestamp = 0; + bwest_str->receiverTimestamp = 0; + + bwest_str->external_bw_info.in_use = 0; + + return 0; +} + +/* This function updates both bottle neck rates */ +/* Parameters: */ +/* rtp_number - value from RTP packet, from NetEq */ +/* frame length - length of signal frame in ms, from iSAC decoder */ +/* send_ts - value in RTP header giving send time in samples */ +/* arr_ts - value given by timeGetTime() time of arrival in samples of packet from NetEq */ +/* pksize - size of packet in bytes, from NetEq */ +/* Index - integer (range 0...23) indicating bottle neck & jitter as estimated by other side */ +/* returns 0 if everything went fine, -1 otherwise */ +int16_t WebRtcIsac_UpdateBandwidthEstimator( + BwEstimatorstr* bwest_str, + const uint16_t rtp_number, + const int32_t frame_length, + const uint32_t send_ts, + const uint32_t arr_ts, + const size_t pksize + /*, const uint16_t Index*/) +{ + float weight = 0.0f; + float curr_bw_inv = 0.0f; + float rec_rtp_rate; + float t_diff_proj; + float arr_ts_diff; + float send_ts_diff; + float arr_time_noise; + float arr_time_noise_abs; + + float delay_correction_factor = 1; + float late_diff = 0.0f; + int immediate_set = 0; + int num_pkts_expected; + + assert(!bwest_str->external_bw_info.in_use); + + // We have to adjust the header-rate if the first packet has a + // frame-size different than the initialized value. + if ( frame_length != bwest_str->prev_frame_length ) + { + bwest_str->rec_header_rate = (float)HEADER_SIZE * 8.0f * + 1000.0f / (float)frame_length; /* bits/s */ + } + + /* UPDATE ESTIMATES ON THIS SIDE */ + /* compute far-side transmission rate */ + rec_rtp_rate = ((float)pksize * 8.0f * 1000.0f / (float)frame_length) + + bwest_str->rec_header_rate; + // rec_rtp_rate packet bits/s + header bits/s + + /* check for timer wrap-around */ + if (arr_ts < bwest_str->prev_rec_arr_ts) + { + bwest_str->prev_rec_arr_ts = arr_ts; + bwest_str->last_update_ts = arr_ts; + bwest_str->last_reduction_ts = arr_ts + 3*FS; + bwest_str->num_pkts_rec = 0; + + /* store frame length */ + bwest_str->prev_frame_length = frame_length; + + /* store far-side transmission rate */ + bwest_str->prev_rec_rtp_rate = rec_rtp_rate; + + /* store far-side RTP time stamp */ + bwest_str->prev_rec_rtp_number = rtp_number; + + return 0; + } + + bwest_str->num_pkts_rec++; + + /* check that it's not one of the first 9 packets */ + if ( bwest_str->count_tot_updates_rec > 0 ) + { + if(bwest_str->in_wait_period > 0 ) + { + bwest_str->in_wait_period--; + } + + bwest_str->inWaitLatePkts -= ((bwest_str->inWaitLatePkts > 0)? 1:0); + send_ts_diff = (float)(send_ts - bwest_str->prev_rec_send_ts); + + if (send_ts_diff <= (16 * frame_length)*2) + //doesn't allow for a dropped packet, not sure necessary to be + // that strict -DH + { + /* if not been updated for a long time, reduce the BN estimate */ + if((uint32_t)(arr_ts - bwest_str->last_update_ts) * + 1000.0f / FS > 3000) + { + //how many frames should have been received since the last + // update if too many have been dropped or there have been + // big delays won't allow this reduction may no longer need + // the send_ts_diff here + num_pkts_expected = (int)(((float)(arr_ts - + bwest_str->last_update_ts) * 1000.0f /(float) FS) / + (float)frame_length); + + if(((float)bwest_str->num_pkts_rec/(float)num_pkts_expected) > + 0.9) + { + float inv_bitrate = (float) pow( 0.99995, + (double)((uint32_t)(arr_ts - + bwest_str->last_reduction_ts)*1000.0f/FS) ); + + if ( inv_bitrate ) + { + bwest_str->rec_bw_inv /= inv_bitrate; + + //precautionary, likely never necessary + if (bwest_str->hsn_detect_snd && + bwest_str->hsn_detect_rec) + { + if (bwest_str->rec_bw_inv > 0.000066f) + { + bwest_str->rec_bw_inv = 0.000066f; + } + } + } + else + { + bwest_str->rec_bw_inv = 1.0f / + (INIT_BN_EST_WB + INIT_HDR_RATE_WB); + } + /* reset time-since-update counter */ + bwest_str->last_reduction_ts = arr_ts; + } + else + //reset here? + { + bwest_str->last_reduction_ts = arr_ts + 3*FS; + bwest_str->last_update_ts = arr_ts; + bwest_str->num_pkts_rec = 0; + } + } + } + else + { + bwest_str->last_reduction_ts = arr_ts + 3*FS; + bwest_str->last_update_ts = arr_ts; + bwest_str->num_pkts_rec = 0; + } + + + /* temporarily speed up adaptation if frame length has changed */ + if ( frame_length != bwest_str->prev_frame_length ) + { + bwest_str->count_tot_updates_rec = 10; + bwest_str->rec_header_rate = (float)HEADER_SIZE * 8.0f * + 1000.0f / (float)frame_length; /* bits/s */ + + bwest_str->rec_bw_inv = 1.0f /((float)bwest_str->rec_bw + + bwest_str->rec_header_rate); + } + + //////////////////////// + arr_ts_diff = (float)(arr_ts - bwest_str->prev_rec_arr_ts); + + if (send_ts_diff > 0 ) + { + late_diff = arr_ts_diff - send_ts_diff; + } + else + { + late_diff = arr_ts_diff - (float)(16 * frame_length); + } + + if((late_diff > 0) && !bwest_str->inWaitLatePkts) + { + bwest_str->numConsecLatePkts++; + bwest_str->consecLatency += late_diff; + } + else + { + bwest_str->numConsecLatePkts = 0; + bwest_str->consecLatency = 0; + } + if(bwest_str->numConsecLatePkts > 50) + { + float latencyMs = bwest_str->consecLatency/(FS/1000); + float averageLatencyMs = latencyMs / bwest_str->numConsecLatePkts; + delay_correction_factor = frame_length / (frame_length + averageLatencyMs); + immediate_set = 1; + bwest_str->inWaitLatePkts = (int16_t)((bwest_str->consecLatency/(FS/1000)) / 30);// + 150; + bwest_str->start_wait_period = arr_ts; + } + /////////////////////////////////////////////// + + + + /* update only if previous packet was not lost */ + if ( rtp_number == bwest_str->prev_rec_rtp_number + 1 ) + { + + + if (!(bwest_str->hsn_detect_snd && bwest_str->hsn_detect_rec)) + { + if ((arr_ts_diff > (float)(16 * frame_length))) + { + //1/2 second + if ((late_diff > 8000.0f) && !bwest_str->in_wait_period) + { + delay_correction_factor = 0.7f; + bwest_str->in_wait_period = 55; + bwest_str->start_wait_period = arr_ts; + immediate_set = 1; + } + //320 ms + else if (late_diff > 5120.0f && !bwest_str->in_wait_period) + { + delay_correction_factor = 0.8f; + immediate_set = 1; + bwest_str->in_wait_period = 44; + bwest_str->start_wait_period = arr_ts; + } + } + } + + + if ((bwest_str->prev_rec_rtp_rate > bwest_str->rec_bw_avg) && + (rec_rtp_rate > bwest_str->rec_bw_avg) && + !bwest_str->in_wait_period) + { + /* test if still in initiation period and increment counter */ + if (bwest_str->count_tot_updates_rec++ > 99) + { + /* constant weight after initiation part */ + weight = 0.01f; + } + else + { + /* weight decreases with number of updates */ + weight = 1.0f / (float) bwest_str->count_tot_updates_rec; + } + /* Bottle Neck Estimation */ + + /* limit outliers */ + /* if more than 25 ms too much */ + if (arr_ts_diff > frame_length * FS/1000 + 400.0f) + { + // in samples, why 25ms?? + arr_ts_diff = frame_length * FS/1000 + 400.0f; + } + if(arr_ts_diff < (frame_length * FS/1000) - 160.0f) + { + /* don't allow it to be less than frame rate - 10 ms */ + arr_ts_diff = (float)frame_length * FS/1000 - 160.0f; + } + + /* compute inverse receiving rate for last packet */ + curr_bw_inv = arr_ts_diff / ((float)(pksize + HEADER_SIZE) * + 8.0f * FS); // (180+35)*8*16000 = 27.5 Mbit.... + + + if(curr_bw_inv < + (1.0f / (MAX_ISAC_BW + bwest_str->rec_header_rate))) + { + // don't allow inv rate to be larger than MAX + curr_bw_inv = (1.0f / + (MAX_ISAC_BW + bwest_str->rec_header_rate)); + } + + /* update bottle neck rate estimate */ + bwest_str->rec_bw_inv = weight * curr_bw_inv + + (1.0f - weight) * bwest_str->rec_bw_inv; + + /* reset time-since-update counter */ + bwest_str->last_update_ts = arr_ts; + bwest_str->last_reduction_ts = arr_ts + 3 * FS; + bwest_str->num_pkts_rec = 0; + + /* Jitter Estimation */ + /* projected difference between arrival times */ + t_diff_proj = ((float)(pksize + HEADER_SIZE) * 8.0f * + 1000.0f) / bwest_str->rec_bw_avg; + + + // difference between projected and actual + // arrival time differences + arr_time_noise = (float)(arr_ts_diff*1000.0f/FS) - + t_diff_proj; + arr_time_noise_abs = (float) fabs( arr_time_noise ); + + /* long term averaged absolute jitter */ + bwest_str->rec_jitter = weight * arr_time_noise_abs + + (1.0f - weight) * bwest_str->rec_jitter; + if (bwest_str->rec_jitter > 10.0f) + { + bwest_str->rec_jitter = 10.0f; + } + /* short term averaged absolute jitter */ + bwest_str->rec_jitter_short_term_abs = 0.05f * + arr_time_noise_abs + 0.95f * + bwest_str->rec_jitter_short_term_abs; + + /* short term averaged jitter */ + bwest_str->rec_jitter_short_term = 0.05f * arr_time_noise + + 0.95f * bwest_str->rec_jitter_short_term; + } + } + } + else + { + // reset time-since-update counter when + // receiving the first 9 packets + bwest_str->last_update_ts = arr_ts; + bwest_str->last_reduction_ts = arr_ts + 3*FS; + bwest_str->num_pkts_rec = 0; + + bwest_str->count_tot_updates_rec++; + } + + /* limit minimum bottle neck rate */ + if (bwest_str->rec_bw_inv > 1.0f / ((float)MIN_ISAC_BW + + bwest_str->rec_header_rate)) + { + bwest_str->rec_bw_inv = 1.0f / ((float)MIN_ISAC_BW + + bwest_str->rec_header_rate); + } + + // limit maximum bitrate + if (bwest_str->rec_bw_inv < 1.0f / ((float)MAX_ISAC_BW + + bwest_str->rec_header_rate)) + { + bwest_str->rec_bw_inv = 1.0f / ((float)MAX_ISAC_BW + + bwest_str->rec_header_rate); + } + + /* store frame length */ + bwest_str->prev_frame_length = frame_length; + + /* store far-side transmission rate */ + bwest_str->prev_rec_rtp_rate = rec_rtp_rate; + + /* store far-side RTP time stamp */ + bwest_str->prev_rec_rtp_number = rtp_number; + + // Replace bwest_str->rec_max_delay by the new + // value (atomic operation) + bwest_str->rec_max_delay = 3.0f * bwest_str->rec_jitter; + + /* store send and arrival time stamp */ + bwest_str->prev_rec_arr_ts = arr_ts ; + bwest_str->prev_rec_send_ts = send_ts; + + /* Replace bwest_str->rec_bw by the new value (atomic operation) */ + bwest_str->rec_bw = (int32_t)(1.0f / bwest_str->rec_bw_inv - + bwest_str->rec_header_rate); + + if (immediate_set) + { + bwest_str->rec_bw = (int32_t) (delay_correction_factor * + (float) bwest_str->rec_bw); + + if (bwest_str->rec_bw < (int32_t) MIN_ISAC_BW) + { + bwest_str->rec_bw = (int32_t) MIN_ISAC_BW; + } + + bwest_str->rec_bw_avg = bwest_str->rec_bw + + bwest_str->rec_header_rate; + + bwest_str->rec_bw_avg_Q = (float) bwest_str->rec_bw; + + bwest_str->rec_jitter_short_term = 0.0f; + + bwest_str->rec_bw_inv = 1.0f / (bwest_str->rec_bw + + bwest_str->rec_header_rate); + + bwest_str->count_tot_updates_rec = 1; + + immediate_set = 0; + bwest_str->consecLatency = 0; + bwest_str->numConsecLatePkts = 0; + } + + return 0; +} + + +/* This function updates the send bottle neck rate */ +/* Index - integer (range 0...23) indicating bottle neck & jitter as estimated by other side */ +/* returns 0 if everything went fine, -1 otherwise */ +int16_t WebRtcIsac_UpdateUplinkBwImpl( + BwEstimatorstr* bwest_str, + int16_t index, + enum IsacSamplingRate encoderSamplingFreq) +{ + assert(!bwest_str->external_bw_info.in_use); + + if((index < 0) || (index > 23)) + { + return -ISAC_RANGE_ERROR_BW_ESTIMATOR; + } + + /* UPDATE ESTIMATES FROM OTHER SIDE */ + if(encoderSamplingFreq == kIsacWideband) + { + if(index > 11) + { + index -= 12; + /* compute the jitter estimate as decoded on the other side */ + bwest_str->send_max_delay_avg = 0.9f * bwest_str->send_max_delay_avg + + 0.1f * (float)MAX_ISAC_MD; + } + else + { + /* compute the jitter estimate as decoded on the other side */ + bwest_str->send_max_delay_avg = 0.9f * bwest_str->send_max_delay_avg + + 0.1f * (float)MIN_ISAC_MD; + } + + /* compute the BN estimate as decoded on the other side */ + bwest_str->send_bw_avg = 0.9f * bwest_str->send_bw_avg + + 0.1f * kQRateTableWb[index]; + } + else + { + /* compute the BN estimate as decoded on the other side */ + bwest_str->send_bw_avg = 0.9f * bwest_str->send_bw_avg + + 0.1f * kQRateTableSwb[index]; + } + + if (bwest_str->send_bw_avg > (float) 28000 && !bwest_str->hsn_detect_snd) + { + bwest_str->num_consec_snt_pkts_over_30k++; + + if (bwest_str->num_consec_snt_pkts_over_30k >= 66) + { + //approx 2 seconds with 30ms frames + bwest_str->hsn_detect_snd = 1; + } + } + else if (!bwest_str->hsn_detect_snd) + { + bwest_str->num_consec_snt_pkts_over_30k = 0; + } + return 0; +} + +// called when there is upper-band bit-stream to update jitter +// statistics. +int16_t WebRtcIsac_UpdateUplinkJitter( + BwEstimatorstr* bwest_str, + int32_t index) +{ + assert(!bwest_str->external_bw_info.in_use); + + if((index < 0) || (index > 23)) + { + return -ISAC_RANGE_ERROR_BW_ESTIMATOR; + } + + if(index > 0) + { + /* compute the jitter estimate as decoded on the other side */ + bwest_str->send_max_delay_avg = 0.9f * bwest_str->send_max_delay_avg + + 0.1f * (float)MAX_ISAC_MD; + } + else + { + /* compute the jitter estimate as decoded on the other side */ + bwest_str->send_max_delay_avg = 0.9f * bwest_str->send_max_delay_avg + + 0.1f * (float)MIN_ISAC_MD; + } + + return 0; +} + + + +// Returns the bandwidth/jitter estimation code (integer 0...23) +// to put in the sending iSAC payload +void +WebRtcIsac_GetDownlinkBwJitIndexImpl( + BwEstimatorstr* bwest_str, + int16_t* bottleneckIndex, + int16_t* jitterInfo, + enum IsacSamplingRate decoderSamplingFreq) +{ + float MaxDelay; + //uint16_t MaxDelayBit; + + float rate; + float r; + float e1, e2; + const float weight = 0.1f; + const float* ptrQuantizationTable; + int16_t addJitterInfo; + int16_t minInd; + int16_t maxInd; + int16_t midInd; + + if (bwest_str->external_bw_info.in_use) { + *bottleneckIndex = bwest_str->external_bw_info.bottleneck_idx; + *jitterInfo = bwest_str->external_bw_info.jitter_info; + return; + } + + /* Get Max Delay Bit */ + /* get unquantized max delay */ + MaxDelay = (float)WebRtcIsac_GetDownlinkMaxDelay(bwest_str); + + if ( ((1.f - weight) * bwest_str->rec_max_delay_avg_Q + weight * + MAX_ISAC_MD - MaxDelay) > (MaxDelay - (1.f-weight) * + bwest_str->rec_max_delay_avg_Q - weight * MIN_ISAC_MD) ) + { + jitterInfo[0] = 0; + /* update quantized average */ + bwest_str->rec_max_delay_avg_Q = + (1.f - weight) * bwest_str->rec_max_delay_avg_Q + weight * + (float)MIN_ISAC_MD; + } + else + { + jitterInfo[0] = 1; + /* update quantized average */ + bwest_str->rec_max_delay_avg_Q = + (1.f-weight) * bwest_str->rec_max_delay_avg_Q + weight * + (float)MAX_ISAC_MD; + } + + // Get unquantized rate. + rate = (float)WebRtcIsac_GetDownlinkBandwidth(bwest_str); + + /* Get Rate Index */ + if(decoderSamplingFreq == kIsacWideband) + { + ptrQuantizationTable = kQRateTableWb; + addJitterInfo = 1; + maxInd = 11; + } + else + { + ptrQuantizationTable = kQRateTableSwb; + addJitterInfo = 0; + maxInd = 23; + } + + minInd = 0; + while(maxInd > minInd + 1) + { + midInd = (maxInd + minInd) >> 1; + if(rate > ptrQuantizationTable[midInd]) + { + minInd = midInd; + } + else + { + maxInd = midInd; + } + } + // Chose the index which gives results an average which is closest + // to rate + r = (1 - weight) * bwest_str->rec_bw_avg_Q - rate; + e1 = weight * ptrQuantizationTable[minInd] + r; + e2 = weight * ptrQuantizationTable[maxInd] + r; + e1 = (e1 > 0)? e1:-e1; + e2 = (e2 > 0)? e2:-e2; + if(e1 < e2) + { + bottleneckIndex[0] = minInd; + } + else + { + bottleneckIndex[0] = maxInd; + } + + bwest_str->rec_bw_avg_Q = (1 - weight) * bwest_str->rec_bw_avg_Q + + weight * ptrQuantizationTable[bottleneckIndex[0]]; + bottleneckIndex[0] += jitterInfo[0] * 12 * addJitterInfo; + + bwest_str->rec_bw_avg = (1 - weight) * bwest_str->rec_bw_avg + weight * + (rate + bwest_str->rec_header_rate); +} + + + +/* get the bottle neck rate from far side to here, as estimated on this side */ +int32_t WebRtcIsac_GetDownlinkBandwidth( const BwEstimatorstr *bwest_str) +{ + int32_t rec_bw; + float jitter_sign; + float bw_adjust; + + assert(!bwest_str->external_bw_info.in_use); + + /* create a value between -1.0 and 1.0 indicating "average sign" of jitter */ + jitter_sign = bwest_str->rec_jitter_short_term / + bwest_str->rec_jitter_short_term_abs; + + /* adjust bw proportionally to negative average jitter sign */ + bw_adjust = 1.0f - jitter_sign * (0.15f + 0.15f * jitter_sign * jitter_sign); + + /* adjust Rate if jitter sign is mostly constant */ + rec_bw = (int32_t)(bwest_str->rec_bw * bw_adjust); + + /* limit range of bottle neck rate */ + if (rec_bw < MIN_ISAC_BW) + { + rec_bw = MIN_ISAC_BW; + } + else if (rec_bw > MAX_ISAC_BW) + { + rec_bw = MAX_ISAC_BW; + } + return rec_bw; +} + +/* Returns the max delay (in ms) */ +int32_t +WebRtcIsac_GetDownlinkMaxDelay(const BwEstimatorstr *bwest_str) +{ + int32_t rec_max_delay; + + assert(!bwest_str->external_bw_info.in_use); + + rec_max_delay = (int32_t)(bwest_str->rec_max_delay); + + /* limit range of jitter estimate */ + if (rec_max_delay < MIN_ISAC_MD) + { + rec_max_delay = MIN_ISAC_MD; + } + else if (rec_max_delay > MAX_ISAC_MD) + { + rec_max_delay = MAX_ISAC_MD; + } + return rec_max_delay; +} + +/* Clamp val to the closed interval [min,max]. */ +static int32_t clamp(int32_t val, int32_t min, int32_t max) { + assert(min <= max); + return val < min ? min : (val > max ? max : val); +} + +int32_t WebRtcIsac_GetUplinkBandwidth(const BwEstimatorstr* bwest_str) { + return bwest_str->external_bw_info.in_use + ? bwest_str->external_bw_info.send_bw_avg + : clamp(bwest_str->send_bw_avg, MIN_ISAC_BW, MAX_ISAC_BW); +} + +int32_t WebRtcIsac_GetUplinkMaxDelay(const BwEstimatorstr* bwest_str) { + return bwest_str->external_bw_info.in_use + ? bwest_str->external_bw_info.send_max_delay_avg + : clamp(bwest_str->send_max_delay_avg, MIN_ISAC_MD, MAX_ISAC_MD); +} + +void WebRtcIsacBw_GetBandwidthInfo(BwEstimatorstr* bwest_str, + enum IsacSamplingRate decoder_sample_rate_hz, + IsacBandwidthInfo* bwinfo) { + assert(!bwest_str->external_bw_info.in_use); + bwinfo->in_use = 1; + bwinfo->send_bw_avg = WebRtcIsac_GetUplinkBandwidth(bwest_str); + bwinfo->send_max_delay_avg = WebRtcIsac_GetUplinkMaxDelay(bwest_str); + WebRtcIsac_GetDownlinkBwJitIndexImpl(bwest_str, &bwinfo->bottleneck_idx, + &bwinfo->jitter_info, + decoder_sample_rate_hz); +} + +void WebRtcIsacBw_SetBandwidthInfo(BwEstimatorstr* bwest_str, + const IsacBandwidthInfo* bwinfo) { + memcpy(&bwest_str->external_bw_info, bwinfo, + sizeof bwest_str->external_bw_info); +} + +/* + * update long-term average bitrate and amount of data in buffer + * returns minimum payload size (bytes) + */ +int WebRtcIsac_GetMinBytes( + RateModel* State, + int StreamSize, /* bytes in bitstream */ + const int FrameSamples, /* samples per frame */ + const double BottleNeck, /* bottle neck rate; excl headers (bps) */ + const double DelayBuildUp, /* max delay from bottleneck buffering (ms) */ + enum ISACBandwidth bandwidth + /*,int16_t frequentLargePackets*/) +{ + double MinRate = 0.0; + int MinBytes; + double TransmissionTime; + int burstInterval = BURST_INTERVAL; + + // first 10 packets @ low rate, then INIT_BURST_LEN packets @ + // fixed rate of INIT_RATE bps + if (State->InitCounter > 0) + { + if (State->InitCounter-- <= INIT_BURST_LEN) + { + if(bandwidth == isac8kHz) + { + MinRate = INIT_RATE_WB; + } + else + { + MinRate = INIT_RATE_SWB; + } + } + else + { + MinRate = 0; + } + } + else + { + /* handle burst */ + if (State->BurstCounter) + { + if (State->StillBuffered < (1.0 - 1.0/BURST_LEN) * DelayBuildUp) + { + /* max bps derived from BottleNeck and DelayBuildUp values */ + MinRate = (1.0 + (FS/1000) * DelayBuildUp / + (double)(BURST_LEN * FrameSamples)) * BottleNeck; + } + else + { + // max bps derived from StillBuffered and DelayBuildUp + // values + MinRate = (1.0 + (FS/1000) * (DelayBuildUp - + State->StillBuffered) / (double)FrameSamples) * BottleNeck; + if (MinRate < 1.04 * BottleNeck) + { + MinRate = 1.04 * BottleNeck; + } + } + State->BurstCounter--; + } + } + + + /* convert rate from bits/second to bytes/packet */ + MinBytes = (int) (MinRate * FrameSamples / (8.0 * FS)); + + /* StreamSize will be adjusted if less than MinBytes */ + if (StreamSize < MinBytes) + { + StreamSize = MinBytes; + } + + /* keep track of when bottle neck was last exceeded by at least 1% */ + if (StreamSize * 8.0 * FS / FrameSamples > 1.01 * BottleNeck) { + if (State->PrevExceed) { + /* bottle_neck exceded twice in a row, decrease ExceedAgo */ + State->ExceedAgo -= /*BURST_INTERVAL*/ burstInterval / (BURST_LEN - 1); + if (State->ExceedAgo < 0) + State->ExceedAgo = 0; + } + else + { + State->ExceedAgo += (FrameSamples * 1000) / FS; /* ms */ + State->PrevExceed = 1; + } + } + else + { + State->PrevExceed = 0; + State->ExceedAgo += (FrameSamples * 1000) / FS; /* ms */ + } + + /* set burst flag if bottle neck not exceeded for long time */ + if ((State->ExceedAgo > burstInterval) && + (State->BurstCounter == 0)) + { + if (State->PrevExceed) + { + State->BurstCounter = BURST_LEN - 1; + } + else + { + State->BurstCounter = BURST_LEN; + } + } + + + /* Update buffer delay */ + TransmissionTime = StreamSize * 8.0 * 1000.0 / BottleNeck; /* ms */ + State->StillBuffered += TransmissionTime; + State->StillBuffered -= (FrameSamples * 1000) / FS; /* ms */ + if (State->StillBuffered < 0.0) + { + State->StillBuffered = 0.0; + } + + return MinBytes; +} + + +/* + * update long-term average bitrate and amount of data in buffer + */ +void WebRtcIsac_UpdateRateModel( + RateModel *State, + int StreamSize, /* bytes in bitstream */ + const int FrameSamples, /* samples per frame */ + const double BottleNeck) /* bottle neck rate; excl headers (bps) */ +{ + double TransmissionTime; + + /* avoid the initial "high-rate" burst */ + State->InitCounter = 0; + + /* Update buffer delay */ + TransmissionTime = StreamSize * 8.0 * 1000.0 / BottleNeck; /* ms */ + State->StillBuffered += TransmissionTime; + State->StillBuffered -= (FrameSamples * 1000) / FS; /* ms */ + if (State->StillBuffered < 0.0) + State->StillBuffered = 0.0; + +} + + +void WebRtcIsac_InitRateModel( + RateModel *State) +{ + State->PrevExceed = 0; /* boolean */ + State->ExceedAgo = 0; /* ms */ + State->BurstCounter = 0; /* packets */ + State->InitCounter = INIT_BURST_LEN + 10; /* packets */ + State->StillBuffered = 1.0; /* ms */ +} + +int WebRtcIsac_GetNewFrameLength( + double bottle_neck, + int current_framesamples) +{ + int new_framesamples; + + const int Thld_20_30 = 20000; + + //const int Thld_30_20 = 30000; + const int Thld_30_20 = 1000000; // disable 20 ms frames + + const int Thld_30_60 = 18000; + //const int Thld_30_60 = 0; // disable 60 ms frames + + const int Thld_60_30 = 27000; + + + new_framesamples = current_framesamples; + + /* find new framelength */ + switch(current_framesamples) { + case 320: + if (bottle_neck < Thld_20_30) + new_framesamples = 480; + break; + case 480: + if (bottle_neck < Thld_30_60) + new_framesamples = 960; + else if (bottle_neck > Thld_30_20) + new_framesamples = 320; + break; + case 960: + if (bottle_neck >= Thld_60_30) + new_framesamples = 480; + break; + } + + return new_framesamples; +} + +double WebRtcIsac_GetSnr( + double bottle_neck, + int framesamples) +{ + double s2nr; + + const double a_20 = -30.0; + const double b_20 = 0.8; + const double c_20 = 0.0; + + const double a_30 = -23.0; + const double b_30 = 0.48; + const double c_30 = 0.0; + + const double a_60 = -23.0; + const double b_60 = 0.53; + const double c_60 = 0.0; + + + /* find new SNR value */ + switch(framesamples) { + case 320: + s2nr = a_20 + b_20 * bottle_neck * 0.001 + c_20 * bottle_neck * + bottle_neck * 0.000001; + break; + case 480: + s2nr = a_30 + b_30 * bottle_neck * 0.001 + c_30 * bottle_neck * + bottle_neck * 0.000001; + break; + case 960: + s2nr = a_60 + b_60 * bottle_neck * 0.001 + c_60 * bottle_neck * + bottle_neck * 0.000001; + break; + default: + s2nr = 0; + } + + return s2nr; + +} diff --git a/webrtc/modules/audio_coding/codecs/isac/main/source/bandwidth_estimator.h b/webrtc/modules/audio_coding/codecs/isac/main/source/bandwidth_estimator.h new file mode 100644 index 0000000..0704337 --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/main/source/bandwidth_estimator.h @@ -0,0 +1,184 @@ +/* + * Copyright (c) 2011 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +/* + * bandwidth_estimator.h + * + * This header file contains the API for the Bandwidth Estimator + * designed for iSAC. + * + */ + +#ifndef WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_SOURCE_BANDWIDTH_ESTIMATOR_H_ +#define WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_SOURCE_BANDWIDTH_ESTIMATOR_H_ + +#include "structs.h" +#include "settings.h" + + +#define MIN_ISAC_BW 10000 +#define MIN_ISAC_BW_LB 10000 +#define MIN_ISAC_BW_UB 25000 + +#define MAX_ISAC_BW 56000 +#define MAX_ISAC_BW_UB 32000 +#define MAX_ISAC_BW_LB 32000 + +#define MIN_ISAC_MD 5 +#define MAX_ISAC_MD 25 + +// assumed header size, in bytes; we don't know the exact number +// (header compression may be used) +#define HEADER_SIZE 35 + +// Initial Frame-Size, in ms, for Wideband & Super-Wideband Mode +#define INIT_FRAME_LEN_WB 60 +#define INIT_FRAME_LEN_SWB 30 + +// Initial Bottleneck Estimate, in bits/sec, for +// Wideband & Super-wideband mode +#define INIT_BN_EST_WB 20e3f +#define INIT_BN_EST_SWB 56e3f + +// Initial Header rate (header rate depends on frame-size), +// in bits/sec, for Wideband & Super-Wideband mode. +#define INIT_HDR_RATE_WB \ + ((float)HEADER_SIZE * 8.0f * 1000.0f / (float)INIT_FRAME_LEN_WB) +#define INIT_HDR_RATE_SWB \ + ((float)HEADER_SIZE * 8.0f * 1000.0f / (float)INIT_FRAME_LEN_SWB) + +// number of packets in a row for a high rate burst +#define BURST_LEN 3 + +// ms, max time between two full bursts +#define BURST_INTERVAL 500 + +// number of packets in a row for initial high rate burst +#define INIT_BURST_LEN 5 + +// bits/s, rate for the first BURST_LEN packets +#define INIT_RATE_WB INIT_BN_EST_WB +#define INIT_RATE_SWB INIT_BN_EST_SWB + + +#if defined(__cplusplus) +extern "C" { +#endif + + /* This function initializes the struct */ + /* to be called before using the struct for anything else */ + /* returns 0 if everything went fine, -1 otherwise */ + int32_t WebRtcIsac_InitBandwidthEstimator( + BwEstimatorstr* bwest_str, + enum IsacSamplingRate encoderSampRate, + enum IsacSamplingRate decoderSampRate); + + /* This function updates the receiving estimate */ + /* Parameters: */ + /* rtp_number - value from RTP packet, from NetEq */ + /* frame length - length of signal frame in ms, from iSAC decoder */ + /* send_ts - value in RTP header giving send time in samples */ + /* arr_ts - value given by timeGetTime() time of arrival in samples of packet from NetEq */ + /* pksize - size of packet in bytes, from NetEq */ + /* Index - integer (range 0...23) indicating bottle neck & jitter as estimated by other side */ + /* returns 0 if everything went fine, -1 otherwise */ + int16_t WebRtcIsac_UpdateBandwidthEstimator( + BwEstimatorstr* bwest_str, + const uint16_t rtp_number, + const int32_t frame_length, + const uint32_t send_ts, + const uint32_t arr_ts, + const size_t pksize); + + /* Update receiving estimates. Used when we only receive BWE index, no iSAC data packet. */ + int16_t WebRtcIsac_UpdateUplinkBwImpl( + BwEstimatorstr* bwest_str, + int16_t Index, + enum IsacSamplingRate encoderSamplingFreq); + + /* Returns the bandwidth/jitter estimation code (integer 0...23) to put in the sending iSAC payload */ + void WebRtcIsac_GetDownlinkBwJitIndexImpl( + BwEstimatorstr* bwest_str, + int16_t* bottleneckIndex, + int16_t* jitterInfo, + enum IsacSamplingRate decoderSamplingFreq); + + /* Returns the bandwidth estimation (in bps) */ + int32_t WebRtcIsac_GetDownlinkBandwidth( + const BwEstimatorstr *bwest_str); + + /* Returns the max delay (in ms) */ + int32_t WebRtcIsac_GetDownlinkMaxDelay( + const BwEstimatorstr *bwest_str); + + /* Returns the bandwidth that iSAC should send with in bps */ + int32_t WebRtcIsac_GetUplinkBandwidth(const BwEstimatorstr* bwest_str); + + /* Returns the max delay value from the other side in ms */ + int32_t WebRtcIsac_GetUplinkMaxDelay( + const BwEstimatorstr *bwest_str); + + /* Fills in an IsacExternalBandwidthInfo struct. */ + void WebRtcIsacBw_GetBandwidthInfo( + BwEstimatorstr* bwest_str, + enum IsacSamplingRate decoder_sample_rate_hz, + IsacBandwidthInfo* bwinfo); + + /* Uses the values from an IsacExternalBandwidthInfo struct. */ + void WebRtcIsacBw_SetBandwidthInfo(BwEstimatorstr* bwest_str, + const IsacBandwidthInfo* bwinfo); + + /* + * update amount of data in bottle neck buffer and burst handling + * returns minimum payload size (bytes) + */ + int WebRtcIsac_GetMinBytes( + RateModel* State, + int StreamSize, /* bytes in bitstream */ + const int FrameLen, /* ms per frame */ + const double BottleNeck, /* bottle neck rate; excl headers (bps) */ + const double DelayBuildUp, /* max delay from bottleneck buffering (ms) */ + enum ISACBandwidth bandwidth + /*,int16_t frequentLargePackets*/); + + /* + * update long-term average bitrate and amount of data in buffer + */ + void WebRtcIsac_UpdateRateModel( + RateModel* State, + int StreamSize, /* bytes in bitstream */ + const int FrameSamples, /* samples per frame */ + const double BottleNeck); /* bottle neck rate; excl headers (bps) */ + + + void WebRtcIsac_InitRateModel( + RateModel *State); + + /* Returns the new framelength value (input argument: bottle_neck) */ + int WebRtcIsac_GetNewFrameLength( + double bottle_neck, + int current_framelength); + + /* Returns the new SNR value (input argument: bottle_neck) */ + double WebRtcIsac_GetSnr( + double bottle_neck, + int new_framelength); + + + int16_t WebRtcIsac_UpdateUplinkJitter( + BwEstimatorstr* bwest_str, + int32_t index); + +#if defined(__cplusplus) +} +#endif + + +#endif /* WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_SOURCE_BANDWIDTH_ESTIMATOR_H_ */ diff --git a/webrtc/modules/audio_coding/codecs/isac/main/source/crc.c b/webrtc/modules/audio_coding/codecs/isac/main/source/crc.c new file mode 100644 index 0000000..ebef595 --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/main/source/crc.c @@ -0,0 +1,110 @@ +/* + * Copyright (c) 2011 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#include "crc.h" +#include +#include "webrtc/common_audio/signal_processing/include/signal_processing_library.h" + +#define POLYNOMIAL 0x04c11db7L + + +static const uint32_t kCrcTable[256] = { + 0, 0x4c11db7, 0x9823b6e, 0xd4326d9, 0x130476dc, 0x17c56b6b, + 0x1a864db2, 0x1e475005, 0x2608edb8, 0x22c9f00f, 0x2f8ad6d6, 0x2b4bcb61, + 0x350c9b64, 0x31cd86d3, 0x3c8ea00a, 0x384fbdbd, 0x4c11db70, 0x48d0c6c7, + 0x4593e01e, 0x4152fda9, 0x5f15adac, 0x5bd4b01b, 0x569796c2, 0x52568b75, + 0x6a1936c8, 0x6ed82b7f, 0x639b0da6, 0x675a1011, 0x791d4014, 0x7ddc5da3, + 0x709f7b7a, 0x745e66cd, 0x9823b6e0, 0x9ce2ab57, 0x91a18d8e, 0x95609039, + 0x8b27c03c, 0x8fe6dd8b, 0x82a5fb52, 0x8664e6e5, 0xbe2b5b58, 0xbaea46ef, + 0xb7a96036, 0xb3687d81, 0xad2f2d84, 0xa9ee3033, 0xa4ad16ea, 0xa06c0b5d, + 0xd4326d90, 0xd0f37027, 0xddb056fe, 0xd9714b49, 0xc7361b4c, 0xc3f706fb, + 0xceb42022, 0xca753d95, 0xf23a8028, 0xf6fb9d9f, 0xfbb8bb46, 0xff79a6f1, + 0xe13ef6f4, 0xe5ffeb43, 0xe8bccd9a, 0xec7dd02d, 0x34867077, 0x30476dc0, + 0x3d044b19, 0x39c556ae, 0x278206ab, 0x23431b1c, 0x2e003dc5, 0x2ac12072, + 0x128e9dcf, 0x164f8078, 0x1b0ca6a1, 0x1fcdbb16, 0x18aeb13, 0x54bf6a4, + 0x808d07d, 0xcc9cdca, 0x7897ab07, 0x7c56b6b0, 0x71159069, 0x75d48dde, + 0x6b93dddb, 0x6f52c06c, 0x6211e6b5, 0x66d0fb02, 0x5e9f46bf, 0x5a5e5b08, + 0x571d7dd1, 0x53dc6066, 0x4d9b3063, 0x495a2dd4, 0x44190b0d, 0x40d816ba, + 0xaca5c697, 0xa864db20, 0xa527fdf9, 0xa1e6e04e, 0xbfa1b04b, 0xbb60adfc, + 0xb6238b25, 0xb2e29692, 0x8aad2b2f, 0x8e6c3698, 0x832f1041, 0x87ee0df6, + 0x99a95df3, 0x9d684044, 0x902b669d, 0x94ea7b2a, 0xe0b41de7, 0xe4750050, + 0xe9362689, 0xedf73b3e, 0xf3b06b3b, 0xf771768c, 0xfa325055, 0xfef34de2, + 0xc6bcf05f, 0xc27dede8, 0xcf3ecb31, 0xcbffd686, 0xd5b88683, 0xd1799b34, + 0xdc3abded, 0xd8fba05a, 0x690ce0ee, 0x6dcdfd59, 0x608edb80, 0x644fc637, + 0x7a089632, 0x7ec98b85, 0x738aad5c, 0x774bb0eb, 0x4f040d56, 0x4bc510e1, + 0x46863638, 0x42472b8f, 0x5c007b8a, 0x58c1663d, 0x558240e4, 0x51435d53, + 0x251d3b9e, 0x21dc2629, 0x2c9f00f0, 0x285e1d47, 0x36194d42, 0x32d850f5, + 0x3f9b762c, 0x3b5a6b9b, 0x315d626, 0x7d4cb91, 0xa97ed48, 0xe56f0ff, + 0x1011a0fa, 0x14d0bd4d, 0x19939b94, 0x1d528623, 0xf12f560e, 0xf5ee4bb9, + 0xf8ad6d60, 0xfc6c70d7, 0xe22b20d2, 0xe6ea3d65, 0xeba91bbc, 0xef68060b, + 0xd727bbb6, 0xd3e6a601, 0xdea580d8, 0xda649d6f, 0xc423cd6a, 0xc0e2d0dd, + 0xcda1f604, 0xc960ebb3, 0xbd3e8d7e, 0xb9ff90c9, 0xb4bcb610, 0xb07daba7, + 0xae3afba2, 0xaafbe615, 0xa7b8c0cc, 0xa379dd7b, 0x9b3660c6, 0x9ff77d71, + 0x92b45ba8, 0x9675461f, 0x8832161a, 0x8cf30bad, 0x81b02d74, 0x857130c3, + 0x5d8a9099, 0x594b8d2e, 0x5408abf7, 0x50c9b640, 0x4e8ee645, 0x4a4ffbf2, + 0x470cdd2b, 0x43cdc09c, 0x7b827d21, 0x7f436096, 0x7200464f, 0x76c15bf8, + 0x68860bfd, 0x6c47164a, 0x61043093, 0x65c52d24, 0x119b4be9, 0x155a565e, + 0x18197087, 0x1cd86d30, 0x29f3d35, 0x65e2082, 0xb1d065b, 0xfdc1bec, + 0x3793a651, 0x3352bbe6, 0x3e119d3f, 0x3ad08088, 0x2497d08d, 0x2056cd3a, + 0x2d15ebe3, 0x29d4f654, 0xc5a92679, 0xc1683bce, 0xcc2b1d17, 0xc8ea00a0, + 0xd6ad50a5, 0xd26c4d12, 0xdf2f6bcb, 0xdbee767c, 0xe3a1cbc1, 0xe760d676, + 0xea23f0af, 0xeee2ed18, 0xf0a5bd1d, 0xf464a0aa, 0xf9278673, 0xfde69bc4, + 0x89b8fd09, 0x8d79e0be, 0x803ac667, 0x84fbdbd0, 0x9abc8bd5, 0x9e7d9662, + 0x933eb0bb, 0x97ffad0c, 0xafb010b1, 0xab710d06, 0xa6322bdf, 0xa2f33668, + 0xbcb4666d, 0xb8757bda, 0xb5365d03, 0xb1f740b4 +}; + + + + +/**************************************************************************** + * WebRtcIsac_GetCrc(...) + * + * This function returns a 32 bit CRC checksum of a bit stream + * + * Input: + * - bitstream : payload bitstream + * - len_bitstream_in_bytes : number of 8-bit words in the bit stream + * + * Output: + * - crc : checksum + * + * Return value : 0 - Ok + * -1 - Error + */ + +int WebRtcIsac_GetCrc(const int16_t* bitstream, + int len_bitstream_in_bytes, + uint32_t* crc) +{ + uint8_t* bitstream_ptr_uw8; + uint32_t crc_state; + int byte_cntr; + int crc_tbl_indx; + + /* Sanity Check. */ + if (bitstream == NULL) { + return -1; + } + /* cast to UWord8 pointer */ + bitstream_ptr_uw8 = (uint8_t *)bitstream; + + /* initialize */ + crc_state = 0xFFFFFFFF; + + for (byte_cntr = 0; byte_cntr < len_bitstream_in_bytes; byte_cntr++) { + crc_tbl_indx = (WEBRTC_SPL_RSHIFT_U32(crc_state, 24) ^ + bitstream_ptr_uw8[byte_cntr]) & 0xFF; + crc_state = (crc_state << 8) ^ kCrcTable[crc_tbl_indx]; + } + + *crc = ~crc_state; + return 0; +} diff --git a/webrtc/modules/audio_coding/codecs/isac/main/source/crc.h b/webrtc/modules/audio_coding/codecs/isac/main/source/crc.h new file mode 100644 index 0000000..09583df --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/main/source/crc.h @@ -0,0 +1,46 @@ +/* + * Copyright (c) 2011 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +/* + * crc.h + * + * Checksum functions + * + */ + +#ifndef WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_SOURCE_CRC_H_ +#define WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_SOURCE_CRC_H_ + +#include "webrtc/typedefs.h" + +/**************************************************************************** + * WebRtcIsac_GetCrc(...) + * + * This function returns a 32 bit CRC checksum of a bit stream + * + * Input: + * - encoded : payload bit stream + * - no_of_word8s : number of 8-bit words in the bit stream + * + * Output: + * - crc : checksum + * + * Return value : 0 - Ok + * -1 - Error + */ + +int WebRtcIsac_GetCrc( + const int16_t* encoded, + int no_of_word8s, + uint32_t* crc); + + + +#endif /* WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_SOURCE_CRC_H_ */ diff --git a/webrtc/modules/audio_coding/codecs/isac/main/source/decode.c b/webrtc/modules/audio_coding/codecs/isac/main/source/decode.c new file mode 100644 index 0000000..e925efb --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/main/source/decode.c @@ -0,0 +1,304 @@ +/* + * Copyright (c) 2012 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +/* + * decode_B.c + * + * This file contains definition of funtions for decoding. + * Decoding of lower-band, including normal-decoding and RCU decoding. + * Decoding of upper-band, including 8-12 kHz, when the bandwidth is + * 0-12 kHz, and 8-16 kHz, when the bandwidth is 0-16 kHz. + * + */ + + +#include "codec.h" +#include "entropy_coding.h" +#include "pitch_estimator.h" +#include "bandwidth_estimator.h" +#include "structs.h" +#include "settings.h" + +#include +#include +#include + + +/* + * function to decode the bitstream + * returns the total number of bytes in the stream + */ +int WebRtcIsac_DecodeLb(const TransformTables* transform_tables, + float* signal_out, ISACLBDecStruct* ISACdecLB_obj, + int16_t* current_framesamples, + int16_t isRCUPayload) { + int k; + int len, err; + int16_t bandwidthInd; + + float LP_dec_float[FRAMESAMPLES_HALF]; + float HP_dec_float[FRAMESAMPLES_HALF]; + + double LPw[FRAMESAMPLES_HALF]; + double HPw[FRAMESAMPLES_HALF]; + double LPw_pf[FRAMESAMPLES_HALF]; + + double lo_filt_coef[(ORDERLO + 1)*SUBFRAMES]; + double hi_filt_coef[(ORDERHI + 1)*SUBFRAMES]; + + double real_f[FRAMESAMPLES_HALF]; + double imag_f[FRAMESAMPLES_HALF]; + + double PitchLags[4]; + double PitchGains[4]; + double AvgPitchGain; + int16_t PitchGains_Q12[4]; + int16_t AvgPitchGain_Q12; + + float gain; + + int frame_nb; /* counter */ + int frame_mode; /* 0 30ms, 1 for 60ms */ + /* Processed_samples: 480 (30, 60 ms). Cannot take other values. */ + + WebRtcIsac_ResetBitstream(&(ISACdecLB_obj->bitstr_obj)); + + len = 0; + + /* Decode framelength and BW estimation - not used, + only for stream pointer*/ + err = WebRtcIsac_DecodeFrameLen(&ISACdecLB_obj->bitstr_obj, + current_framesamples); + if (err < 0) { + return err; + } + + /* Frame_mode: + * 0: indicates 30 ms frame (480 samples) + * 1: indicates 60 ms frame (960 samples) */ + frame_mode = *current_framesamples / MAX_FRAMESAMPLES; + + err = WebRtcIsac_DecodeSendBW(&ISACdecLB_obj->bitstr_obj, &bandwidthInd); + if (err < 0) { + return err; + } + + /* One loop if it's one frame (20 or 30ms), 2 loops if 2 frames + bundled together (60ms). */ + for (frame_nb = 0; frame_nb <= frame_mode; frame_nb++) { + /* Decode & de-quantize pitch parameters */ + err = WebRtcIsac_DecodePitchGain(&ISACdecLB_obj->bitstr_obj, + PitchGains_Q12); + if (err < 0) { + return err; + } + + err = WebRtcIsac_DecodePitchLag(&ISACdecLB_obj->bitstr_obj, PitchGains_Q12, + PitchLags); + if (err < 0) { + return err; + } + + AvgPitchGain_Q12 = (PitchGains_Q12[0] + PitchGains_Q12[1] + + PitchGains_Q12[2] + PitchGains_Q12[3]) >> 2; + + /* Decode & de-quantize filter coefficients. */ + err = WebRtcIsac_DecodeLpc(&ISACdecLB_obj->bitstr_obj, lo_filt_coef, + hi_filt_coef); + if (err < 0) { + return err; + } + /* Decode & de-quantize spectrum. */ + len = WebRtcIsac_DecodeSpec(&ISACdecLB_obj->bitstr_obj, AvgPitchGain_Q12, + kIsacLowerBand, real_f, imag_f); + if (len < 0) { + return len; + } + + /* Inverse transform. */ + WebRtcIsac_Spec2time(transform_tables, real_f, imag_f, LPw, HPw, + &ISACdecLB_obj->fftstr_obj); + + /* Convert PitchGains back to float for pitchfilter_post */ + for (k = 0; k < 4; k++) { + PitchGains[k] = ((float)PitchGains_Q12[k]) / 4096; + } + if (isRCUPayload) { + for (k = 0; k < 240; k++) { + LPw[k] *= RCU_TRANSCODING_SCALE_INVERSE; + HPw[k] *= RCU_TRANSCODING_SCALE_INVERSE; + } + } + + /* Inverse pitch filter. */ + WebRtcIsac_PitchfilterPost(LPw, LPw_pf, &ISACdecLB_obj->pitchfiltstr_obj, + PitchLags, PitchGains); + /* Convert AvgPitchGain back to float for computation of gain. */ + AvgPitchGain = ((float)AvgPitchGain_Q12) / 4096; + gain = 1.0f - 0.45f * (float)AvgPitchGain; + + for (k = 0; k < FRAMESAMPLES_HALF; k++) { + /* Reduce gain to compensate for pitch enhancer. */ + LPw_pf[k] *= gain; + } + + if (isRCUPayload) { + for (k = 0; k < FRAMESAMPLES_HALF; k++) { + /* Compensation for transcoding gain changes. */ + LPw_pf[k] *= RCU_TRANSCODING_SCALE; + HPw[k] *= RCU_TRANSCODING_SCALE; + } + } + /* Perceptual post-filtering (using normalized lattice filter). */ + WebRtcIsac_NormLatticeFilterAr( + ORDERLO, ISACdecLB_obj->maskfiltstr_obj.PostStateLoF, + (ISACdecLB_obj->maskfiltstr_obj).PostStateLoG, LPw_pf, lo_filt_coef, + LP_dec_float); + WebRtcIsac_NormLatticeFilterAr( + ORDERHI, ISACdecLB_obj->maskfiltstr_obj.PostStateHiF, + (ISACdecLB_obj->maskfiltstr_obj).PostStateHiG, HPw, hi_filt_coef, + HP_dec_float); + + /* Recombine the 2 bands. */ + WebRtcIsac_FilterAndCombineFloat(LP_dec_float, HP_dec_float, + signal_out + frame_nb * FRAMESAMPLES, + &ISACdecLB_obj->postfiltbankstr_obj); + } + return len; +} + + +/* + * This decode function is called when the codec is operating in 16 kHz + * bandwidth to decode the upperband, i.e. 8-16 kHz. + * + * Contrary to lower-band, the upper-band (8-16 kHz) is not split in + * frequency, but split to 12 sub-frames, i.e. twice as lower-band. + */ +int WebRtcIsac_DecodeUb16(const TransformTables* transform_tables, + float* signal_out, ISACUBDecStruct* ISACdecUB_obj, + int16_t isRCUPayload) { + int len, err; + + double halfFrameFirst[FRAMESAMPLES_HALF]; + double halfFrameSecond[FRAMESAMPLES_HALF]; + + double percepFilterParam[(UB_LPC_ORDER + 1) * (SUBFRAMES << 1) + + (UB_LPC_ORDER + 1)]; + + double real_f[FRAMESAMPLES_HALF]; + double imag_f[FRAMESAMPLES_HALF]; + const int16_t kAveragePitchGain = 0; /* No pitch-gain for upper-band. */ + len = 0; + + /* Decode & de-quantize filter coefficients. */ + memset(percepFilterParam, 0, sizeof(percepFilterParam)); + err = WebRtcIsac_DecodeInterpolLpcUb(&ISACdecUB_obj->bitstr_obj, + percepFilterParam, isac16kHz); + if (err < 0) { + return err; + } + + /* Decode & de-quantize spectrum. */ + len = WebRtcIsac_DecodeSpec(&ISACdecUB_obj->bitstr_obj, kAveragePitchGain, + kIsacUpperBand16, real_f, imag_f); + if (len < 0) { + return len; + } + if (isRCUPayload) { + int n; + for (n = 0; n < 240; n++) { + real_f[n] *= RCU_TRANSCODING_SCALE_UB_INVERSE; + imag_f[n] *= RCU_TRANSCODING_SCALE_UB_INVERSE; + } + } + /* Inverse transform. */ + WebRtcIsac_Spec2time(transform_tables, + real_f, imag_f, halfFrameFirst, halfFrameSecond, + &ISACdecUB_obj->fftstr_obj); + + /* Perceptual post-filtering (using normalized lattice filter). */ + WebRtcIsac_NormLatticeFilterAr( + UB_LPC_ORDER, ISACdecUB_obj->maskfiltstr_obj.PostStateLoF, + (ISACdecUB_obj->maskfiltstr_obj).PostStateLoG, halfFrameFirst, + &percepFilterParam[(UB_LPC_ORDER + 1)], signal_out); + + WebRtcIsac_NormLatticeFilterAr( + UB_LPC_ORDER, ISACdecUB_obj->maskfiltstr_obj.PostStateLoF, + (ISACdecUB_obj->maskfiltstr_obj).PostStateLoG, halfFrameSecond, + &percepFilterParam[(UB_LPC_ORDER + 1) * SUBFRAMES + (UB_LPC_ORDER + 1)], + &signal_out[FRAMESAMPLES_HALF]); + + return len; +} + +/* + * This decode function is called when the codec operates at 0-12 kHz + * bandwidth to decode the upperband, i.e. 8-12 kHz. + * + * At the encoder the upper-band is split into two band, 8-12 kHz & 12-16 + * kHz, and only 8-12 kHz is encoded. At the decoder, 8-12 kHz band is + * reconstructed and 12-16 kHz replaced with zeros. Then two bands + * are combined, to reconstruct the upperband 8-16 kHz. + */ +int WebRtcIsac_DecodeUb12(const TransformTables* transform_tables, + float* signal_out, ISACUBDecStruct* ISACdecUB_obj, + int16_t isRCUPayload) { + int len, err; + + float LP_dec_float[FRAMESAMPLES_HALF]; + float HP_dec_float[FRAMESAMPLES_HALF]; + + double LPw[FRAMESAMPLES_HALF]; + double HPw[FRAMESAMPLES_HALF]; + + double percepFilterParam[(UB_LPC_ORDER + 1)*SUBFRAMES]; + + double real_f[FRAMESAMPLES_HALF]; + double imag_f[FRAMESAMPLES_HALF]; + const int16_t kAveragePitchGain = 0; /* No pitch-gain for upper-band. */ + len = 0; + + /* Decode & dequantize filter coefficients. */ + err = WebRtcIsac_DecodeInterpolLpcUb(&ISACdecUB_obj->bitstr_obj, + percepFilterParam, isac12kHz); + if (err < 0) { + return err; + } + + /* Decode & de-quantize spectrum. */ + len = WebRtcIsac_DecodeSpec(&ISACdecUB_obj->bitstr_obj, kAveragePitchGain, + kIsacUpperBand12, real_f, imag_f); + if (len < 0) { + return len; + } + + if (isRCUPayload) { + int n; + for (n = 0; n < 240; n++) { + real_f[n] *= RCU_TRANSCODING_SCALE_UB_INVERSE; + imag_f[n] *= RCU_TRANSCODING_SCALE_UB_INVERSE; + } + } + /* Inverse transform. */ + WebRtcIsac_Spec2time(transform_tables, + real_f, imag_f, LPw, HPw, &ISACdecUB_obj->fftstr_obj); + /* perceptual post-filtering (using normalized lattice filter) */ + WebRtcIsac_NormLatticeFilterAr(UB_LPC_ORDER, + ISACdecUB_obj->maskfiltstr_obj.PostStateLoF, + (ISACdecUB_obj->maskfiltstr_obj).PostStateLoG, + LPw, percepFilterParam, LP_dec_float); + /* Zero for 12-16 kHz. */ + memset(HP_dec_float, 0, sizeof(float) * (FRAMESAMPLES_HALF)); + /* Recombine the 2 bands. */ + WebRtcIsac_FilterAndCombineFloat(HP_dec_float, LP_dec_float, signal_out, + &ISACdecUB_obj->postfiltbankstr_obj); + return len; +} diff --git a/webrtc/modules/audio_coding/codecs/isac/main/source/decode_bwe.c b/webrtc/modules/audio_coding/codecs/isac/main/source/decode_bwe.c new file mode 100644 index 0000000..019cc89 --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/main/source/decode_bwe.c @@ -0,0 +1,89 @@ +/* + * Copyright (c) 2011 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#include "structs.h" +#include "bandwidth_estimator.h" +#include "entropy_coding.h" +#include "codec.h" + + +int +WebRtcIsac_EstimateBandwidth( + BwEstimatorstr* bwest_str, + Bitstr* streamdata, + size_t packet_size, + uint16_t rtp_seq_number, + uint32_t send_ts, + uint32_t arr_ts, + enum IsacSamplingRate encoderSampRate, + enum IsacSamplingRate decoderSampRate) +{ + int16_t index; + int16_t frame_samples; + uint32_t sendTimestampIn16kHz; + uint32_t arrivalTimestampIn16kHz; + uint32_t diffSendTime; + uint32_t diffArrivalTime; + int err; + + /* decode framelength and BW estimation */ + err = WebRtcIsac_DecodeFrameLen(streamdata, &frame_samples); + if(err < 0) // error check + { + return err; + } + err = WebRtcIsac_DecodeSendBW(streamdata, &index); + if(err < 0) // error check + { + return err; + } + + /* UPDATE ESTIMATES FROM OTHER SIDE */ + err = WebRtcIsac_UpdateUplinkBwImpl(bwest_str, index, encoderSampRate); + if(err < 0) + { + return err; + } + + // We like BWE to work at 16 kHz sampling rate, + // therefore, we have to change the timestamps accordingly. + // translate the send timestamp if required + diffSendTime = (uint32_t)((uint32_t)send_ts - + (uint32_t)bwest_str->senderTimestamp); + bwest_str->senderTimestamp = send_ts; + + diffArrivalTime = (uint32_t)((uint32_t)arr_ts - + (uint32_t)bwest_str->receiverTimestamp); + bwest_str->receiverTimestamp = arr_ts; + + if(decoderSampRate == kIsacSuperWideband) + { + diffArrivalTime = (uint32_t)diffArrivalTime >> 1; + diffSendTime = (uint32_t)diffSendTime >> 1; + } + + // arrival timestamp in 16 kHz + arrivalTimestampIn16kHz = (uint32_t)((uint32_t) + bwest_str->prev_rec_arr_ts + (uint32_t)diffArrivalTime); + // send timestamp in 16 kHz + sendTimestampIn16kHz = (uint32_t)((uint32_t) + bwest_str->prev_rec_send_ts + (uint32_t)diffSendTime); + + err = WebRtcIsac_UpdateBandwidthEstimator(bwest_str, rtp_seq_number, + (frame_samples * 1000) / FS, sendTimestampIn16kHz, + arrivalTimestampIn16kHz, packet_size); + // error check + if(err < 0) + { + return err; + } + + return 0; +} diff --git a/webrtc/modules/audio_coding/codecs/isac/main/source/encode.c b/webrtc/modules/audio_coding/codecs/isac/main/source/encode.c new file mode 100644 index 0000000..3f1912b --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/main/source/encode.c @@ -0,0 +1,1258 @@ +/* + * Copyright (c) 2012 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +/* + * encode.c + * + * This file contains definition of funtions for encoding. + * Decoding of upper-band, including 8-12 kHz, when the bandwidth is + * 0-12 kHz, and 8-16 kHz, when the bandwidth is 0-16 kHz. + * + */ + +#include +#include +#include + +#include "structs.h" +#include "codec.h" +#include "pitch_estimator.h" +#include "entropy_coding.h" +#include "arith_routines.h" +#include "pitch_gain_tables.h" +#include "pitch_lag_tables.h" +#include "spectrum_ar_model_tables.h" +#include "lpc_tables.h" +#include "lpc_analysis.h" +#include "bandwidth_estimator.h" +#include "lpc_shape_swb12_tables.h" +#include "lpc_shape_swb16_tables.h" +#include "lpc_gain_swb_tables.h" + + +#define UB_LOOKAHEAD 24 + + +/* + Rate allocation tables of lower and upper-band bottleneck for + 12kHz & 16kHz bandwidth. + + 12 kHz bandwidth + ----------------- + The overall bottleneck of the coder is between 38 kbps and 45 kbps. We have + considered 7 enteries, uniformly distributed in this interval, i.e. 38, + 39.17, 40.33, 41.5, 42.67, 43.83 and 45. For every entery, the lower-band + and the upper-band bottlenecks are specified in + 'kLowerBandBitRate12' and 'kUpperBandBitRate12' + tables, respectively. E.g. the overall rate of 41.5 kbps corresponts to a + bottleneck of 31 kbps for lower-band and 27 kbps for upper-band. Given an + overall bottleneck of the codec, we use linear interpolation to get + lower-band and upper-band bottlenecks. + + 16 kHz bandwidth + ----------------- + The overall bottleneck of the coder is between 50 kbps and 56 kbps. We have + considered 7 enteries, uniformly distributed in this interval, i.e. 50, 51.2, + 52.4, 53.6, 54.8 and 56. For every entery, the lower-band and the upper-band + bottlenecks are specified in 'kLowerBandBitRate16' and + 'kUpperBandBitRate16' tables, respectively. E.g. the overall rate + of 53.6 kbps corresponts to a bottleneck of 32 kbps for lower-band and 30 + kbps for upper-band. Given an overall bottleneck of the codec, we use linear + interpolation to get lower-band and upper-band bottlenecks. + + */ + +/* 38 39.17 40.33 41.5 42.67 43.83 45 */ +static const int16_t kLowerBandBitRate12[7] = { + 29000, 30000, 30000, 31000, 31000, 32000, 32000 }; +static const int16_t kUpperBandBitRate12[7] = { + 25000, 25000, 27000, 27000, 29000, 29000, 32000 }; + +/* 50 51.2 52.4 53.6 54.8 56 */ +static const int16_t kLowerBandBitRate16[6] = { + 31000, 31000, 32000, 32000, 32000, 32000 }; +static const int16_t kUpperBandBitRate16[6] = { + 28000, 29000, 29000, 30000, 31000, 32000 }; + +/****************************************************************************** + * WebRtcIsac_RateAllocation() + * Internal function to perform a rate-allocation for upper and lower-band, + * given a total rate. + * + * Input: + * - inRateBitPerSec : a total bottleneck in bits/sec. + * + * Output: + * - rateLBBitPerSec : a bottleneck allocated to the lower-band + * in bits/sec. + * - rateUBBitPerSec : a bottleneck allocated to the upper-band + * in bits/sec. + * + * Return value : 0 if rate allocation has been successful. + * -1 if failed to allocate rates. + */ + +int16_t WebRtcIsac_RateAllocation(int32_t inRateBitPerSec, + double* rateLBBitPerSec, + double* rateUBBitPerSec, + enum ISACBandwidth* bandwidthKHz) { + int16_t idx; + double idxD; + double idxErr; + if (inRateBitPerSec < 38000) { + /* If the given overall bottleneck is less than 38000 then + * then codec has to operate in wideband mode, i.e. 8 kHz + * bandwidth. */ + *rateLBBitPerSec = (int16_t)((inRateBitPerSec > 32000) ? + 32000 : inRateBitPerSec); + *rateUBBitPerSec = 0; + *bandwidthKHz = isac8kHz; + } else if ((inRateBitPerSec >= 38000) && (inRateBitPerSec < 50000)) { + /* At a bottleneck between 38 and 50 kbps the codec is operating + * at 12 kHz bandwidth. Using xxxBandBitRate12[] to calculates + * upper/lower bottleneck */ + + /* Find the bottlenecks by linear interpolation, + * step is (45000 - 38000)/6.0 we use the inverse of it. */ + const double stepSizeInv = 8.5714286e-4; + idxD = (inRateBitPerSec - 38000) * stepSizeInv; + idx = (idxD >= 6) ? 6 : ((int16_t)idxD); + idxErr = idxD - idx; + *rateLBBitPerSec = kLowerBandBitRate12[idx]; + *rateUBBitPerSec = kUpperBandBitRate12[idx]; + + if (idx < 6) { + *rateLBBitPerSec += (int16_t)( + idxErr * (kLowerBandBitRate12[idx + 1] - kLowerBandBitRate12[idx])); + *rateUBBitPerSec += (int16_t)( + idxErr * (kUpperBandBitRate12[idx + 1] - kUpperBandBitRate12[idx])); + } + *bandwidthKHz = isac12kHz; + } else if ((inRateBitPerSec >= 50000) && (inRateBitPerSec <= 56000)) { + /* A bottleneck between 50 and 56 kbps corresponds to bandwidth + * of 16 kHz. Using xxxBandBitRate16[] to calculates + * upper/lower bottleneck. */ + + /* Find the bottlenecks by linear interpolation + * step is (56000 - 50000)/5 we use the inverse of it. */ + const double stepSizeInv = 8.3333333e-4; + idxD = (inRateBitPerSec - 50000) * stepSizeInv; + idx = (idxD >= 5) ? 5 : ((int16_t)idxD); + idxErr = idxD - idx; + *rateLBBitPerSec = kLowerBandBitRate16[idx]; + *rateUBBitPerSec = kUpperBandBitRate16[idx]; + + if (idx < 5) { + *rateLBBitPerSec += (int16_t)(idxErr * + (kLowerBandBitRate16[idx + 1] - + kLowerBandBitRate16[idx])); + + *rateUBBitPerSec += (int16_t)(idxErr * + (kUpperBandBitRate16[idx + 1] - + kUpperBandBitRate16[idx])); + } + *bandwidthKHz = isac16kHz; + } else { + /* Out-of-range botlteneck value. */ + return -1; + } + + /* limit the values. */ + *rateLBBitPerSec = (*rateLBBitPerSec > 32000) ? 32000 : *rateLBBitPerSec; + *rateUBBitPerSec = (*rateUBBitPerSec > 32000) ? 32000 : *rateUBBitPerSec; + return 0; +} + + +void WebRtcIsac_ResetBitstream(Bitstr* bit_stream) { + bit_stream->W_upper = 0xFFFFFFFF; + bit_stream->stream_index = 0; + bit_stream->streamval = 0; +} + +int WebRtcIsac_EncodeLb(const TransformTables* transform_tables, + float* in, ISACLBEncStruct* ISACencLB_obj, + int16_t codingMode, + int16_t bottleneckIndex) { + int stream_length = 0; + int err; + int k; + int iterCntr; + + double lofilt_coef[(ORDERLO + 1)*SUBFRAMES]; + double hifilt_coef[(ORDERHI + 1)*SUBFRAMES]; + float LP[FRAMESAMPLES_HALF]; + float HP[FRAMESAMPLES_HALF]; + + double LP_lookahead[FRAMESAMPLES_HALF]; + double HP_lookahead[FRAMESAMPLES_HALF]; + double LP_lookahead_pf[FRAMESAMPLES_HALF + QLOOKAHEAD]; + double LPw[FRAMESAMPLES_HALF]; + + double HPw[FRAMESAMPLES_HALF]; + double LPw_pf[FRAMESAMPLES_HALF]; + int16_t fre[FRAMESAMPLES_HALF]; /* Q7 */ + int16_t fim[FRAMESAMPLES_HALF]; /* Q7 */ + + double PitchLags[4]; + double PitchGains[4]; + int16_t PitchGains_Q12[4]; + int16_t AvgPitchGain_Q12; + + int frame_mode; /* 0 for 30ms, 1 for 60ms */ + int status = 0; + int my_index; + transcode_obj transcodingParam; + double bytesLeftSpecCoding; + uint16_t payloadLimitBytes; + + /* Copy new frame-length and bottleneck rate only for the first 10 ms data */ + if (ISACencLB_obj->buffer_index == 0) { + /* Set the framelength for the next packet. */ + ISACencLB_obj->current_framesamples = ISACencLB_obj->new_framelength; + } + /* 'frame_mode' is 0 (30 ms) or 1 (60 ms). */ + frame_mode = ISACencLB_obj->current_framesamples / MAX_FRAMESAMPLES; + + /* buffer speech samples (by 10ms packet) until the frame-length */ + /* is reached (30 or 60 ms). */ + /*****************************************************************/ + + /* fill the buffer with 10ms input data */ + for (k = 0; k < FRAMESAMPLES_10ms; k++) { + ISACencLB_obj->data_buffer_float[k + ISACencLB_obj->buffer_index] = in[k]; + } + + /* If buffersize is not equal to current framesize then increase index + * and return. We do no encoding untill we have enough audio. */ + if (ISACencLB_obj->buffer_index + FRAMESAMPLES_10ms != FRAMESAMPLES) { + ISACencLB_obj->buffer_index += FRAMESAMPLES_10ms; + return 0; + } + /* If buffer reached the right size, reset index and continue with + * encoding the frame. */ + ISACencLB_obj->buffer_index = 0; + + /* End of buffer function. */ + /**************************/ + + /* Encoding */ + /************/ + + if (frame_mode == 0 || ISACencLB_obj->frame_nb == 0) { + /* This is to avoid Linux warnings until we change 'int' to 'Word32' + * at all places. */ + int intVar; + /* reset bitstream */ + WebRtcIsac_ResetBitstream(&(ISACencLB_obj->bitstr_obj)); + + if ((codingMode == 0) && (frame_mode == 0) && + (ISACencLB_obj->enforceFrameSize == 0)) { + ISACencLB_obj->new_framelength = WebRtcIsac_GetNewFrameLength( + ISACencLB_obj->bottleneck, ISACencLB_obj->current_framesamples); + } + + ISACencLB_obj->s2nr = WebRtcIsac_GetSnr( + ISACencLB_obj->bottleneck, ISACencLB_obj->current_framesamples); + + /* Encode frame length. */ + status = WebRtcIsac_EncodeFrameLen( + ISACencLB_obj->current_framesamples, &ISACencLB_obj->bitstr_obj); + if (status < 0) { + /* Wrong frame size. */ + return status; + } + /* Save framelength for multiple packets memory. */ + ISACencLB_obj->SaveEnc_obj.framelength = + ISACencLB_obj->current_framesamples; + + /* To be used for Redundant Coding. */ + ISACencLB_obj->lastBWIdx = bottleneckIndex; + intVar = (int)bottleneckIndex; + WebRtcIsac_EncodeReceiveBw(&intVar, &ISACencLB_obj->bitstr_obj); + } + + /* Split signal in two bands. */ + WebRtcIsac_SplitAndFilterFloat(ISACencLB_obj->data_buffer_float, LP, HP, + LP_lookahead, HP_lookahead, + &ISACencLB_obj->prefiltbankstr_obj); + + /* estimate pitch parameters and pitch-filter lookahead signal */ + WebRtcIsac_PitchAnalysis(LP_lookahead, LP_lookahead_pf, + &ISACencLB_obj->pitchanalysisstr_obj, PitchLags, + PitchGains); + + /* Encode in FIX Q12. */ + + /* Convert PitchGain to Fixed point. */ + for (k = 0; k < PITCH_SUBFRAMES; k++) { + PitchGains_Q12[k] = (int16_t)(PitchGains[k] * 4096.0); + } + + /* Set where to store data in multiple packets memory. */ + if (frame_mode == 0 || ISACencLB_obj->frame_nb == 0) { + ISACencLB_obj->SaveEnc_obj.startIdx = 0; + } else { + ISACencLB_obj->SaveEnc_obj.startIdx = 1; + } + + /* Quantize & encode pitch parameters. */ + WebRtcIsac_EncodePitchGain(PitchGains_Q12, &ISACencLB_obj->bitstr_obj, + &ISACencLB_obj->SaveEnc_obj); + WebRtcIsac_EncodePitchLag(PitchLags, PitchGains_Q12, + &ISACencLB_obj->bitstr_obj, + &ISACencLB_obj->SaveEnc_obj); + + AvgPitchGain_Q12 = (PitchGains_Q12[0] + PitchGains_Q12[1] + + PitchGains_Q12[2] + PitchGains_Q12[3]) >> 2; + + /* Find coefficients for perceptual pre-filters. */ + WebRtcIsac_GetLpcCoefLb(LP_lookahead_pf, HP_lookahead, + &ISACencLB_obj->maskfiltstr_obj, ISACencLB_obj->s2nr, + PitchGains_Q12, lofilt_coef, hifilt_coef); + + /* Code LPC model and shape - gains not quantized yet. */ + WebRtcIsac_EncodeLpcLb(lofilt_coef, hifilt_coef, &ISACencLB_obj->bitstr_obj, + &ISACencLB_obj->SaveEnc_obj); + + /* Convert PitchGains back to FLOAT for pitchfilter_pre. */ + for (k = 0; k < 4; k++) { + PitchGains[k] = ((float)PitchGains_Q12[k]) / 4096; + } + + /* Store the state of arithmetic coder before coding LPC gains. */ + transcodingParam.W_upper = ISACencLB_obj->bitstr_obj.W_upper; + transcodingParam.stream_index = ISACencLB_obj->bitstr_obj.stream_index; + transcodingParam.streamval = ISACencLB_obj->bitstr_obj.streamval; + transcodingParam.stream[0] = + ISACencLB_obj->bitstr_obj.stream[ISACencLB_obj->bitstr_obj.stream_index - + 2]; + transcodingParam.stream[1] = + ISACencLB_obj->bitstr_obj.stream[ISACencLB_obj->bitstr_obj.stream_index - + 1]; + transcodingParam.stream[2] = + ISACencLB_obj->bitstr_obj.stream[ISACencLB_obj->bitstr_obj.stream_index]; + + /* Store LPC Gains before encoding them. */ + for (k = 0; k < SUBFRAMES; k++) { + transcodingParam.loFiltGain[k] = lofilt_coef[(LPC_LOBAND_ORDER + 1) * k]; + transcodingParam.hiFiltGain[k] = hifilt_coef[(LPC_HIBAND_ORDER + 1) * k]; + } + + /* Code gains */ + WebRtcIsac_EncodeLpcGainLb(lofilt_coef, hifilt_coef, + &ISACencLB_obj->bitstr_obj, + &ISACencLB_obj->SaveEnc_obj); + + /* Get the correct value for the payload limit and calculate the + * number of bytes left for coding the spectrum. */ + if ((frame_mode == 1) && (ISACencLB_obj->frame_nb == 0)) { + /* It is a 60ms and we are in the first 30ms then the limit at + * this point should be half of the assigned value. */ + payloadLimitBytes = ISACencLB_obj->payloadLimitBytes60 >> 1; + } else if (frame_mode == 0) { + /* It is a 30ms frame */ + /* Subract 3 because termination process may add 3 bytes. */ + payloadLimitBytes = ISACencLB_obj->payloadLimitBytes30 - 3; + } else { + /* This is the second half of a 60ms frame. */ + /* Subract 3 because termination process may add 3 bytes. */ + payloadLimitBytes = ISACencLB_obj->payloadLimitBytes60 - 3; + } + bytesLeftSpecCoding = payloadLimitBytes - transcodingParam.stream_index; + + /* Perceptual pre-filtering (using normalized lattice filter). */ + /* Low-band filtering. */ + WebRtcIsac_NormLatticeFilterMa(ORDERLO, + ISACencLB_obj->maskfiltstr_obj.PreStateLoF, + ISACencLB_obj->maskfiltstr_obj.PreStateLoG, + LP, lofilt_coef, LPw); + /* High-band filtering. */ + WebRtcIsac_NormLatticeFilterMa(ORDERHI, + ISACencLB_obj->maskfiltstr_obj.PreStateHiF, + ISACencLB_obj->maskfiltstr_obj.PreStateHiG, + HP, hifilt_coef, HPw); + /* Pitch filter. */ + WebRtcIsac_PitchfilterPre(LPw, LPw_pf, &ISACencLB_obj->pitchfiltstr_obj, + PitchLags, PitchGains); + /* Transform */ + WebRtcIsac_Time2Spec(transform_tables, + LPw_pf, HPw, fre, fim, &ISACencLB_obj->fftstr_obj); + + /* Save data for multiple packets memory. */ + my_index = ISACencLB_obj->SaveEnc_obj.startIdx * FRAMESAMPLES_HALF; + memcpy(&ISACencLB_obj->SaveEnc_obj.fre[my_index], fre, sizeof(fre)); + memcpy(&ISACencLB_obj->SaveEnc_obj.fim[my_index], fim, sizeof(fim)); + + ISACencLB_obj->SaveEnc_obj.AvgPitchGain[ISACencLB_obj->SaveEnc_obj.startIdx] = + AvgPitchGain_Q12; + + /* Quantization and loss-less coding. */ + err = WebRtcIsac_EncodeSpec(fre, fim, AvgPitchGain_Q12, kIsacLowerBand, + &ISACencLB_obj->bitstr_obj); + if ((err < 0) && (err != -ISAC_DISALLOWED_BITSTREAM_LENGTH)) { + /* There has been an error but it was not too large payload + (we can cure too large payload). */ + if (frame_mode == 1 && ISACencLB_obj->frame_nb == 1) { + /* If this is the second 30ms of a 60ms frame reset + this such that in the next call encoder starts fresh. */ + ISACencLB_obj->frame_nb = 0; + } + return err; + } + iterCntr = 0; + while ((ISACencLB_obj->bitstr_obj.stream_index > payloadLimitBytes) || + (err == -ISAC_DISALLOWED_BITSTREAM_LENGTH)) { + double bytesSpecCoderUsed; + double transcodeScale; + + if (iterCntr >= MAX_PAYLOAD_LIMIT_ITERATION) { + /* We were not able to limit the payload size */ + if ((frame_mode == 1) && (ISACencLB_obj->frame_nb == 0)) { + /* This was the first 30ms of a 60ms frame. Although + the payload is larger than it should be but we let + the second 30ms be encoded. Maybe together we + won't exceed the limit. */ + ISACencLB_obj->frame_nb = 1; + return 0; + } else if ((frame_mode == 1) && (ISACencLB_obj->frame_nb == 1)) { + ISACencLB_obj->frame_nb = 0; + } + + if (err != -ISAC_DISALLOWED_BITSTREAM_LENGTH) { + return -ISAC_PAYLOAD_LARGER_THAN_LIMIT; + } else { + return status; + } + } + + if (err == -ISAC_DISALLOWED_BITSTREAM_LENGTH) { + bytesSpecCoderUsed = STREAM_SIZE_MAX; + /* Being conservative */ + transcodeScale = bytesLeftSpecCoding / bytesSpecCoderUsed * 0.5; + } else { + bytesSpecCoderUsed = ISACencLB_obj->bitstr_obj.stream_index - + transcodingParam.stream_index; + transcodeScale = bytesLeftSpecCoding / bytesSpecCoderUsed; + } + + /* To be safe, we reduce the scale depending on + the number of iterations. */ + transcodeScale *= (1.0 - (0.9 * (double)iterCntr / + (double)MAX_PAYLOAD_LIMIT_ITERATION)); + + /* Scale the LPC Gains. */ + for (k = 0; k < SUBFRAMES; k++) { + lofilt_coef[(LPC_LOBAND_ORDER + 1) * k] = + transcodingParam.loFiltGain[k] * transcodeScale; + hifilt_coef[(LPC_HIBAND_ORDER + 1) * k] = + transcodingParam.hiFiltGain[k] * transcodeScale; + transcodingParam.loFiltGain[k] = lofilt_coef[(LPC_LOBAND_ORDER + 1) * k]; + transcodingParam.hiFiltGain[k] = hifilt_coef[(LPC_HIBAND_ORDER + 1) * k]; + } + + /* Scale DFT coefficients. */ + for (k = 0; k < FRAMESAMPLES_HALF; k++) { + fre[k] = (int16_t)(fre[k] * transcodeScale); + fim[k] = (int16_t)(fim[k] * transcodeScale); + } + + /* Save data for multiple packets memory. */ + my_index = ISACencLB_obj->SaveEnc_obj.startIdx * FRAMESAMPLES_HALF; + memcpy(&ISACencLB_obj->SaveEnc_obj.fre[my_index], fre, sizeof(fre)); + memcpy(&ISACencLB_obj->SaveEnc_obj.fim[my_index], fim, sizeof(fim)); + + /* Re-store the state of arithmetic coder before coding LPC gains. */ + ISACencLB_obj->bitstr_obj.W_upper = transcodingParam.W_upper; + ISACencLB_obj->bitstr_obj.stream_index = transcodingParam.stream_index; + ISACencLB_obj->bitstr_obj.streamval = transcodingParam.streamval; + ISACencLB_obj->bitstr_obj.stream[transcodingParam.stream_index - 2] = + transcodingParam.stream[0]; + ISACencLB_obj->bitstr_obj.stream[transcodingParam.stream_index - 1] = + transcodingParam.stream[1]; + ISACencLB_obj->bitstr_obj.stream[transcodingParam.stream_index] = + transcodingParam.stream[2]; + + /* Code gains. */ + WebRtcIsac_EncodeLpcGainLb(lofilt_coef, hifilt_coef, + &ISACencLB_obj->bitstr_obj, + &ISACencLB_obj->SaveEnc_obj); + + /* Update the number of bytes left for encoding the spectrum. */ + bytesLeftSpecCoding = payloadLimitBytes - transcodingParam.stream_index; + + /* Encode the spectrum. */ + err = WebRtcIsac_EncodeSpec(fre, fim, AvgPitchGain_Q12, kIsacLowerBand, + &ISACencLB_obj->bitstr_obj); + + if ((err < 0) && (err != -ISAC_DISALLOWED_BITSTREAM_LENGTH)) { + /* There has been an error but it was not too large + payload (we can cure too large payload). */ + if (frame_mode == 1 && ISACencLB_obj->frame_nb == 1) { + /* If this is the second 30 ms of a 60 ms frame reset + this such that in the next call encoder starts fresh. */ + ISACencLB_obj->frame_nb = 0; + } + return err; + } + iterCntr++; + } + + /* If 60 ms frame-size and just processed the first 30 ms, */ + /* go back to main function to buffer the other 30 ms speech frame. */ + if (frame_mode == 1) { + if (ISACencLB_obj->frame_nb == 0) { + ISACencLB_obj->frame_nb = 1; + return 0; + } else if (ISACencLB_obj->frame_nb == 1) { + ISACencLB_obj->frame_nb = 0; + /* Also update the frame-length for next packet, + in Adaptive mode only. */ + if (codingMode == 0 && (ISACencLB_obj->enforceFrameSize == 0)) { + ISACencLB_obj->new_framelength = + WebRtcIsac_GetNewFrameLength(ISACencLB_obj->bottleneck, + ISACencLB_obj->current_framesamples); + } + } + } else { + ISACencLB_obj->frame_nb = 0; + } + + /* Complete arithmetic coding. */ + stream_length = WebRtcIsac_EncTerminate(&ISACencLB_obj->bitstr_obj); + return stream_length; +} + + + +static int LimitPayloadUb(ISACUBEncStruct* ISACencUB_obj, + uint16_t payloadLimitBytes, + double bytesLeftSpecCoding, + transcode_obj* transcodingParam, + int16_t* fre, int16_t* fim, + double* lpcGains, enum ISACBand band, int status) { + + int iterCntr = 0; + int k; + double bytesSpecCoderUsed; + double transcodeScale; + const int16_t kAveragePitchGain = 0.0; + + do { + if (iterCntr >= MAX_PAYLOAD_LIMIT_ITERATION) { + /* We were not able to limit the payload size. */ + return -ISAC_PAYLOAD_LARGER_THAN_LIMIT; + } + + if (status == -ISAC_DISALLOWED_BITSTREAM_LENGTH) { + bytesSpecCoderUsed = STREAM_SIZE_MAX; + /* Being conservative. */ + transcodeScale = bytesLeftSpecCoding / bytesSpecCoderUsed * 0.5; + } else { + bytesSpecCoderUsed = ISACencUB_obj->bitstr_obj.stream_index - + transcodingParam->stream_index; + transcodeScale = bytesLeftSpecCoding / bytesSpecCoderUsed; + } + + /* To be safe, we reduce the scale depending on the + number of iterations. */ + transcodeScale *= (1.0 - (0.9 * (double)iterCntr / + (double)MAX_PAYLOAD_LIMIT_ITERATION)); + + /* Scale the LPC Gains. */ + if (band == kIsacUpperBand16) { + /* Two sets of coefficients if 16 kHz. */ + for (k = 0; k < SUBFRAMES; k++) { + transcodingParam->loFiltGain[k] *= transcodeScale; + transcodingParam->hiFiltGain[k] *= transcodeScale; + } + } else { + /* One sets of coefficients if 12 kHz. */ + for (k = 0; k < SUBFRAMES; k++) { + transcodingParam->loFiltGain[k] *= transcodeScale; + } + } + + /* Scale DFT coefficients. */ + for (k = 0; k < FRAMESAMPLES_HALF; k++) { + fre[k] = (int16_t)(fre[k] * transcodeScale + 0.5); + fim[k] = (int16_t)(fim[k] * transcodeScale + 0.5); + } + /* Store FFT coefficients for multiple encoding. */ + memcpy(ISACencUB_obj->SaveEnc_obj.realFFT, fre, + sizeof(ISACencUB_obj->SaveEnc_obj.realFFT)); + memcpy(ISACencUB_obj->SaveEnc_obj.imagFFT, fim, + sizeof(ISACencUB_obj->SaveEnc_obj.imagFFT)); + + /* Store the state of arithmetic coder before coding LPC gains */ + ISACencUB_obj->bitstr_obj.W_upper = transcodingParam->W_upper; + ISACencUB_obj->bitstr_obj.stream_index = transcodingParam->stream_index; + ISACencUB_obj->bitstr_obj.streamval = transcodingParam->streamval; + ISACencUB_obj->bitstr_obj.stream[transcodingParam->stream_index - 2] = + transcodingParam->stream[0]; + ISACencUB_obj->bitstr_obj.stream[transcodingParam->stream_index - 1] = + transcodingParam->stream[1]; + ISACencUB_obj->bitstr_obj.stream[transcodingParam->stream_index] = + transcodingParam->stream[2]; + + /* Store the gains for multiple encoding. */ + memcpy(ISACencUB_obj->SaveEnc_obj.lpcGain, lpcGains, + SUBFRAMES * sizeof(double)); + /* Entropy Code lpc-gains, indices are stored for a later use.*/ + WebRtcIsac_EncodeLpcGainUb(transcodingParam->loFiltGain, + &ISACencUB_obj->bitstr_obj, + ISACencUB_obj->SaveEnc_obj.lpcGainIndex); + + /* If 16kHz should do one more set. */ + if (band == kIsacUpperBand16) { + /* Store the gains for multiple encoding. */ + memcpy(&ISACencUB_obj->SaveEnc_obj.lpcGain[SUBFRAMES], + &lpcGains[SUBFRAMES], SUBFRAMES * sizeof(double)); + /* Entropy Code lpc-gains, indices are stored for a later use.*/ + WebRtcIsac_EncodeLpcGainUb( + transcodingParam->hiFiltGain, &ISACencUB_obj->bitstr_obj, + &ISACencUB_obj->SaveEnc_obj.lpcGainIndex[SUBFRAMES]); + } + + /* Update the number of bytes left for encoding the spectrum. */ + bytesLeftSpecCoding = payloadLimitBytes - + ISACencUB_obj->bitstr_obj.stream_index; + + /* Save the bit-stream object at this point for FEC. */ + memcpy(&ISACencUB_obj->SaveEnc_obj.bitStreamObj, + &ISACencUB_obj->bitstr_obj, sizeof(Bitstr)); + + /* Encode the spectrum. */ + status = WebRtcIsac_EncodeSpec(fre, fim, kAveragePitchGain, + band, &ISACencUB_obj->bitstr_obj); + if ((status < 0) && (status != -ISAC_DISALLOWED_BITSTREAM_LENGTH)) { + /* There has been an error but it was not too large payload + (we can cure too large payload). */ + return status; + } + iterCntr++; + } while ((ISACencUB_obj->bitstr_obj.stream_index > payloadLimitBytes) || + (status == -ISAC_DISALLOWED_BITSTREAM_LENGTH)); + return 0; +} + +int WebRtcIsac_EncodeUb16(const TransformTables* transform_tables, + float* in, ISACUBEncStruct* ISACencUB_obj, + int32_t jitterInfo) { + int err; + int k; + + double lpcVecs[UB_LPC_ORDER * UB16_LPC_VEC_PER_FRAME]; + double percepFilterParams[(1 + UB_LPC_ORDER) * (SUBFRAMES << 1) + + (1 + UB_LPC_ORDER)]; + + double LP_lookahead[FRAMESAMPLES]; + int16_t fre[FRAMESAMPLES_HALF]; /* Q7 */ + int16_t fim[FRAMESAMPLES_HALF]; /* Q7 */ + + int status = 0; + + double varscale[2]; + double corr[SUBFRAMES << 1][UB_LPC_ORDER + 1]; + double lpcGains[SUBFRAMES << 1]; + transcode_obj transcodingParam; + uint16_t payloadLimitBytes; + double s2nr; + const int16_t kAveragePitchGain = 0.0; + int bytesLeftSpecCoding; + + /* Buffer speech samples (by 10ms packet) until the frame-length is */ + /* reached (30 ms). */ + /*********************************************************************/ + + /* fill the buffer with 10ms input data */ + memcpy(&ISACencUB_obj->data_buffer_float[ISACencUB_obj->buffer_index], in, + FRAMESAMPLES_10ms * sizeof(float)); + + /* If buffer size is not equal to current frame-size, and end of file is + * not reached yet, we don't do encoding unless we have the whole frame. */ + if (ISACencUB_obj->buffer_index + FRAMESAMPLES_10ms < FRAMESAMPLES) { + ISACencUB_obj->buffer_index += FRAMESAMPLES_10ms; + return 0; + } + + /* End of buffer function. */ + /**************************/ + + /* Encoding */ + /************/ + + /* Reset bit-stream */ + WebRtcIsac_ResetBitstream(&(ISACencUB_obj->bitstr_obj)); + + /* Encoding of bandwidth information. */ + WebRtcIsac_EncodeJitterInfo(jitterInfo, &ISACencUB_obj->bitstr_obj); + + status = WebRtcIsac_EncodeBandwidth(isac16kHz, &ISACencUB_obj->bitstr_obj); + if (status < 0) { + return status; + } + + s2nr = WebRtcIsac_GetSnr(ISACencUB_obj->bottleneck, FRAMESAMPLES); + + memcpy(lpcVecs, ISACencUB_obj->lastLPCVec, UB_LPC_ORDER * sizeof(double)); + + for (k = 0; k < FRAMESAMPLES; k++) { + LP_lookahead[k] = ISACencUB_obj->data_buffer_float[UB_LOOKAHEAD + k]; + } + + /* Find coefficients for perceptual pre-filters. */ + WebRtcIsac_GetLpcCoefUb(LP_lookahead, &ISACencUB_obj->maskfiltstr_obj, + &lpcVecs[UB_LPC_ORDER], corr, varscale, isac16kHz); + + memcpy(ISACencUB_obj->lastLPCVec, + &lpcVecs[(UB16_LPC_VEC_PER_FRAME - 1) * (UB_LPC_ORDER)], + sizeof(double) * UB_LPC_ORDER); + + /* Code LPC model and shape - gains not quantized yet. */ + WebRtcIsac_EncodeLpcUB(lpcVecs, &ISACencUB_obj->bitstr_obj, + percepFilterParams, isac16kHz, + &ISACencUB_obj->SaveEnc_obj); + + /* the first set of lpc parameters are from the last sub-frame of + * the previous frame. so we don't care about them. */ + WebRtcIsac_GetLpcGain(s2nr, &percepFilterParams[UB_LPC_ORDER + 1], + (SUBFRAMES << 1), lpcGains, corr, varscale); + + /* Store the state of arithmetic coder before coding LPC gains */ + transcodingParam.stream_index = ISACencUB_obj->bitstr_obj.stream_index; + transcodingParam.W_upper = ISACencUB_obj->bitstr_obj.W_upper; + transcodingParam.streamval = ISACencUB_obj->bitstr_obj.streamval; + transcodingParam.stream[0] = + ISACencUB_obj->bitstr_obj.stream[ISACencUB_obj->bitstr_obj.stream_index - + 2]; + transcodingParam.stream[1] = + ISACencUB_obj->bitstr_obj.stream[ISACencUB_obj->bitstr_obj.stream_index - + 1]; + transcodingParam.stream[2] = + ISACencUB_obj->bitstr_obj.stream[ISACencUB_obj->bitstr_obj.stream_index]; + + /* Store LPC Gains before encoding them. */ + for (k = 0; k < SUBFRAMES; k++) { + transcodingParam.loFiltGain[k] = lpcGains[k]; + transcodingParam.hiFiltGain[k] = lpcGains[SUBFRAMES + k]; + } + + /* Store the gains for multiple encoding. */ + memcpy(ISACencUB_obj->SaveEnc_obj.lpcGain, lpcGains, + (SUBFRAMES << 1) * sizeof(double)); + + WebRtcIsac_EncodeLpcGainUb(lpcGains, &ISACencUB_obj->bitstr_obj, + ISACencUB_obj->SaveEnc_obj.lpcGainIndex); + WebRtcIsac_EncodeLpcGainUb( + &lpcGains[SUBFRAMES], &ISACencUB_obj->bitstr_obj, + &ISACencUB_obj->SaveEnc_obj.lpcGainIndex[SUBFRAMES]); + + /* Get the correct value for the payload limit and calculate the number of + bytes left for coding the spectrum. It is a 30ms frame + Subract 3 because termination process may add 3 bytes */ + payloadLimitBytes = ISACencUB_obj->maxPayloadSizeBytes - + ISACencUB_obj->numBytesUsed - 3; + bytesLeftSpecCoding = payloadLimitBytes - + ISACencUB_obj->bitstr_obj.stream_index; + + for (k = 0; k < (SUBFRAMES << 1); k++) { + percepFilterParams[k * (UB_LPC_ORDER + 1) + (UB_LPC_ORDER + 1)] = + lpcGains[k]; + } + + /* LPC filtering (using normalized lattice filter), */ + /* first half-frame. */ + WebRtcIsac_NormLatticeFilterMa(UB_LPC_ORDER, + ISACencUB_obj->maskfiltstr_obj.PreStateLoF, + ISACencUB_obj->maskfiltstr_obj.PreStateLoG, + &ISACencUB_obj->data_buffer_float[0], + &percepFilterParams[UB_LPC_ORDER + 1], + &LP_lookahead[0]); + + /* Second half-frame filtering. */ + WebRtcIsac_NormLatticeFilterMa( + UB_LPC_ORDER, ISACencUB_obj->maskfiltstr_obj.PreStateLoF, + ISACencUB_obj->maskfiltstr_obj.PreStateLoG, + &ISACencUB_obj->data_buffer_float[FRAMESAMPLES_HALF], + &percepFilterParams[(UB_LPC_ORDER + 1) + SUBFRAMES * (UB_LPC_ORDER + 1)], + &LP_lookahead[FRAMESAMPLES_HALF]); + + WebRtcIsac_Time2Spec(transform_tables, + &LP_lookahead[0], &LP_lookahead[FRAMESAMPLES_HALF], + fre, fim, &ISACencUB_obj->fftstr_obj); + + /* Store FFT coefficients for multiple encoding. */ + memcpy(ISACencUB_obj->SaveEnc_obj.realFFT, fre, sizeof(fre)); + memcpy(ISACencUB_obj->SaveEnc_obj.imagFFT, fim, sizeof(fim)); + + /* Prepare the audio buffer for the next packet + * move the last 3 ms to the beginning of the buffer. */ + memcpy(ISACencUB_obj->data_buffer_float, + &ISACencUB_obj->data_buffer_float[FRAMESAMPLES], + LB_TOTAL_DELAY_SAMPLES * sizeof(float)); + /* start writing with 3 ms delay to compensate for the delay + * of the lower-band. */ + ISACencUB_obj->buffer_index = LB_TOTAL_DELAY_SAMPLES; + + /* Save the bit-stream object at this point for FEC. */ + memcpy(&ISACencUB_obj->SaveEnc_obj.bitStreamObj, &ISACencUB_obj->bitstr_obj, + sizeof(Bitstr)); + + /* Qantization and lossless coding */ + /* Note that there is no pitch-gain for this band so kAveragePitchGain = 0 + * is passed to the function. In fact, the function ignores the 3rd parameter + * for this band. */ + err = WebRtcIsac_EncodeSpec(fre, fim, kAveragePitchGain, kIsacUpperBand16, + &ISACencUB_obj->bitstr_obj); + if ((err < 0) && (err != -ISAC_DISALLOWED_BITSTREAM_LENGTH)) { + return err; + } + + if ((ISACencUB_obj->bitstr_obj.stream_index > payloadLimitBytes) || + (err == -ISAC_DISALLOWED_BITSTREAM_LENGTH)) { + err = LimitPayloadUb(ISACencUB_obj, payloadLimitBytes, bytesLeftSpecCoding, + &transcodingParam, fre, fim, lpcGains, + kIsacUpperBand16, err); + } + if (err < 0) { + return err; + } + /* Complete arithmetic coding. */ + return WebRtcIsac_EncTerminate(&ISACencUB_obj->bitstr_obj); +} + + +int WebRtcIsac_EncodeUb12(const TransformTables* transform_tables, + float* in, ISACUBEncStruct* ISACencUB_obj, + int32_t jitterInfo) { + int err; + int k; + + double lpcVecs[UB_LPC_ORDER * UB_LPC_VEC_PER_FRAME]; + + double percepFilterParams[(1 + UB_LPC_ORDER) * SUBFRAMES]; + float LP[FRAMESAMPLES_HALF]; + float HP[FRAMESAMPLES_HALF]; + + double LP_lookahead[FRAMESAMPLES_HALF]; + double HP_lookahead[FRAMESAMPLES_HALF]; + double LPw[FRAMESAMPLES_HALF]; + + double HPw[FRAMESAMPLES_HALF]; + int16_t fre[FRAMESAMPLES_HALF]; /* Q7 */ + int16_t fim[FRAMESAMPLES_HALF]; /* Q7 */ + + int status = 0; + + double varscale[1]; + + double corr[UB_LPC_GAIN_DIM][UB_LPC_ORDER + 1]; + double lpcGains[SUBFRAMES]; + transcode_obj transcodingParam; + uint16_t payloadLimitBytes; + double s2nr; + const int16_t kAveragePitchGain = 0.0; + double bytesLeftSpecCoding; + + /* Buffer speech samples (by 10ms packet) until the framelength is */ + /* reached (30 ms). */ + /********************************************************************/ + + /* Fill the buffer with 10ms input data. */ + memcpy(&ISACencUB_obj->data_buffer_float[ISACencUB_obj->buffer_index], in, + FRAMESAMPLES_10ms * sizeof(float)); + + /* if buffer-size is not equal to current frame-size then increase the + index and return. We do the encoding when we have enough audio. */ + if (ISACencUB_obj->buffer_index + FRAMESAMPLES_10ms < FRAMESAMPLES) { + ISACencUB_obj->buffer_index += FRAMESAMPLES_10ms; + return 0; + } + /* If buffer reached the right size, reset index and continue + with encoding the frame */ + ISACencUB_obj->buffer_index = 0; + + /* End of buffer function */ + /**************************/ + + /* Encoding */ + /************/ + + /* Reset bit-stream. */ + WebRtcIsac_ResetBitstream(&(ISACencUB_obj->bitstr_obj)); + + /* Encoding bandwidth information. */ + WebRtcIsac_EncodeJitterInfo(jitterInfo, &ISACencUB_obj->bitstr_obj); + status = WebRtcIsac_EncodeBandwidth(isac12kHz, &ISACencUB_obj->bitstr_obj); + if (status < 0) { + return status; + } + + s2nr = WebRtcIsac_GetSnr(ISACencUB_obj->bottleneck, FRAMESAMPLES); + + /* Split signal in two bands. */ + WebRtcIsac_SplitAndFilterFloat(ISACencUB_obj->data_buffer_float, HP, LP, + HP_lookahead, LP_lookahead, + &ISACencUB_obj->prefiltbankstr_obj); + + /* Find coefficients for perceptual pre-filters. */ + WebRtcIsac_GetLpcCoefUb(LP_lookahead, &ISACencUB_obj->maskfiltstr_obj, + lpcVecs, corr, varscale, isac12kHz); + + /* Code LPC model and shape - gains not quantized yet. */ + WebRtcIsac_EncodeLpcUB(lpcVecs, &ISACencUB_obj->bitstr_obj, + percepFilterParams, isac12kHz, + &ISACencUB_obj->SaveEnc_obj); + + WebRtcIsac_GetLpcGain(s2nr, percepFilterParams, SUBFRAMES, lpcGains, corr, + varscale); + + /* Store the state of arithmetic coder before coding LPC gains. */ + transcodingParam.W_upper = ISACencUB_obj->bitstr_obj.W_upper; + transcodingParam.stream_index = ISACencUB_obj->bitstr_obj.stream_index; + transcodingParam.streamval = ISACencUB_obj->bitstr_obj.streamval; + transcodingParam.stream[0] = + ISACencUB_obj->bitstr_obj.stream[ISACencUB_obj->bitstr_obj.stream_index - + 2]; + transcodingParam.stream[1] = + ISACencUB_obj->bitstr_obj.stream[ISACencUB_obj->bitstr_obj.stream_index - + 1]; + transcodingParam.stream[2] = + ISACencUB_obj->bitstr_obj.stream[ISACencUB_obj->bitstr_obj.stream_index]; + + /* Store LPC Gains before encoding them. */ + for (k = 0; k < SUBFRAMES; k++) { + transcodingParam.loFiltGain[k] = lpcGains[k]; + } + + /* Store the gains for multiple encoding. */ + memcpy(ISACencUB_obj->SaveEnc_obj.lpcGain, lpcGains, SUBFRAMES * + sizeof(double)); + + WebRtcIsac_EncodeLpcGainUb(lpcGains, &ISACencUB_obj->bitstr_obj, + ISACencUB_obj->SaveEnc_obj.lpcGainIndex); + + for (k = 0; k < SUBFRAMES; k++) { + percepFilterParams[k * (UB_LPC_ORDER + 1)] = lpcGains[k]; + } + + /* perceptual pre-filtering (using normalized lattice filter) */ + /* low-band filtering */ + WebRtcIsac_NormLatticeFilterMa(UB_LPC_ORDER, + ISACencUB_obj->maskfiltstr_obj.PreStateLoF, + ISACencUB_obj->maskfiltstr_obj.PreStateLoG, LP, + percepFilterParams, LPw); + + /* Get the correct value for the payload limit and calculate the number + of bytes left for coding the spectrum. It is a 30ms frame Subract 3 + because termination process may add 3 bytes */ + payloadLimitBytes = ISACencUB_obj->maxPayloadSizeBytes - + ISACencUB_obj->numBytesUsed - 3; + bytesLeftSpecCoding = payloadLimitBytes - + ISACencUB_obj->bitstr_obj.stream_index; + + memset(HPw, 0, sizeof(HPw)); + + /* Transform */ + WebRtcIsac_Time2Spec(transform_tables, + LPw, HPw, fre, fim, &ISACencUB_obj->fftstr_obj); + + /* Store FFT coefficients for multiple encoding. */ + memcpy(ISACencUB_obj->SaveEnc_obj.realFFT, fre, + sizeof(ISACencUB_obj->SaveEnc_obj.realFFT)); + memcpy(ISACencUB_obj->SaveEnc_obj.imagFFT, fim, + sizeof(ISACencUB_obj->SaveEnc_obj.imagFFT)); + + /* Save the bit-stream object at this point for FEC. */ + memcpy(&ISACencUB_obj->SaveEnc_obj.bitStreamObj, + &ISACencUB_obj->bitstr_obj, sizeof(Bitstr)); + + /* Quantization and loss-less coding */ + /* The 4th parameter to this function is pitch-gain, which is only used + * when encoding 0-8 kHz band, and irrelevant in this function, therefore, + * we insert zero here. */ + err = WebRtcIsac_EncodeSpec(fre, fim, kAveragePitchGain, kIsacUpperBand12, + &ISACencUB_obj->bitstr_obj); + if ((err < 0) && (err != -ISAC_DISALLOWED_BITSTREAM_LENGTH)) { + /* There has been an error but it was not too large + payload (we can cure too large payload) */ + return err; + } + + if ((ISACencUB_obj->bitstr_obj.stream_index > payloadLimitBytes) || + (err == -ISAC_DISALLOWED_BITSTREAM_LENGTH)) { + err = LimitPayloadUb(ISACencUB_obj, payloadLimitBytes, bytesLeftSpecCoding, + &transcodingParam, fre, fim, lpcGains, + kIsacUpperBand12, err); + } + if (err < 0) { + return err; + } + /* Complete arithmetic coding. */ + return WebRtcIsac_EncTerminate(&ISACencUB_obj->bitstr_obj); +} + + + + + + +/* This function is used to create a new bit-stream with new BWE. + The same data as previously encoded with the function WebRtcIsac_Encoder(). + The data needed is taken from the structure, where it was stored + when calling the encoder. */ + +int WebRtcIsac_EncodeStoredDataLb(const IsacSaveEncoderData* ISACSavedEnc_obj, + Bitstr* ISACBitStr_obj, int BWnumber, + float scale) { + int ii; + int status; + int BWno = BWnumber; + + const uint16_t* WebRtcIsac_kQPitchGainCdf_ptr[1]; + const uint16_t** cdf; + + double tmpLPCcoeffs_lo[(ORDERLO + 1)*SUBFRAMES * 2]; + double tmpLPCcoeffs_hi[(ORDERHI + 1)*SUBFRAMES * 2]; + int tmpLPCindex_g[12 * 2]; + int16_t tmp_fre[FRAMESAMPLES], tmp_fim[FRAMESAMPLES]; + const int kModel = 0; + + /* Sanity Check - possible values for BWnumber is 0 - 23. */ + if ((BWnumber < 0) || (BWnumber > 23)) { + return -ISAC_RANGE_ERROR_BW_ESTIMATOR; + } + + /* Reset bit-stream. */ + WebRtcIsac_ResetBitstream(ISACBitStr_obj); + + /* Encode frame length */ + status = WebRtcIsac_EncodeFrameLen(ISACSavedEnc_obj->framelength, + ISACBitStr_obj); + if (status < 0) { + /* Wrong frame size. */ + return status; + } + + /* Transcoding */ + if ((scale > 0.0) && (scale < 1.0)) { + /* Compensate LPC gain. */ + for (ii = 0; + ii < ((ORDERLO + 1)* SUBFRAMES * (1 + ISACSavedEnc_obj->startIdx)); + ii++) { + tmpLPCcoeffs_lo[ii] = scale * ISACSavedEnc_obj->LPCcoeffs_lo[ii]; + } + for (ii = 0; + ii < ((ORDERHI + 1) * SUBFRAMES * (1 + ISACSavedEnc_obj->startIdx)); + ii++) { + tmpLPCcoeffs_hi[ii] = scale * ISACSavedEnc_obj->LPCcoeffs_hi[ii]; + } + /* Scale DFT. */ + for (ii = 0; + ii < (FRAMESAMPLES_HALF * (1 + ISACSavedEnc_obj->startIdx)); + ii++) { + tmp_fre[ii] = (int16_t)((scale) * (float)ISACSavedEnc_obj->fre[ii]); + tmp_fim[ii] = (int16_t)((scale) * (float)ISACSavedEnc_obj->fim[ii]); + } + } else { + for (ii = 0; + ii < (KLT_ORDER_GAIN * (1 + ISACSavedEnc_obj->startIdx)); + ii++) { + tmpLPCindex_g[ii] = ISACSavedEnc_obj->LPCindex_g[ii]; + } + for (ii = 0; + ii < (FRAMESAMPLES_HALF * (1 + ISACSavedEnc_obj->startIdx)); + ii++) { + tmp_fre[ii] = ISACSavedEnc_obj->fre[ii]; + tmp_fim[ii] = ISACSavedEnc_obj->fim[ii]; + } + } + + /* Encode bandwidth estimate. */ + WebRtcIsac_EncodeReceiveBw(&BWno, ISACBitStr_obj); + + /* Loop over number of 30 msec */ + for (ii = 0; ii <= ISACSavedEnc_obj->startIdx; ii++) { + /* Encode pitch gains. */ + *WebRtcIsac_kQPitchGainCdf_ptr = WebRtcIsac_kQPitchGainCdf; + WebRtcIsac_EncHistMulti(ISACBitStr_obj, + &ISACSavedEnc_obj->pitchGain_index[ii], + WebRtcIsac_kQPitchGainCdf_ptr, 1); + + /* Entropy coding of quantization pitch lags */ + /* Voicing classification. */ + if (ISACSavedEnc_obj->meanGain[ii] < 0.2) { + cdf = WebRtcIsac_kQPitchLagCdfPtrLo; + } else if (ISACSavedEnc_obj->meanGain[ii] < 0.4) { + cdf = WebRtcIsac_kQPitchLagCdfPtrMid; + } else { + cdf = WebRtcIsac_kQPitchLagCdfPtrHi; + } + WebRtcIsac_EncHistMulti(ISACBitStr_obj, + &ISACSavedEnc_obj->pitchIndex[PITCH_SUBFRAMES * ii], + cdf, PITCH_SUBFRAMES); + + /* LPC */ + /* Only one model exists. The entropy coding is done only for backward + * compatibility. */ + WebRtcIsac_EncHistMulti(ISACBitStr_obj, &kModel, + WebRtcIsac_kQKltModelCdfPtr, 1); + /* Entropy coding of quantization indices - LPC shape only. */ + WebRtcIsac_EncHistMulti(ISACBitStr_obj, + &ISACSavedEnc_obj->LPCindex_s[KLT_ORDER_SHAPE * ii], + WebRtcIsac_kQKltCdfPtrShape, + KLT_ORDER_SHAPE); + + /* If transcoding, get new LPC gain indices */ + if (scale < 1.0) { + WebRtcIsac_TranscodeLPCCoef( + &tmpLPCcoeffs_lo[(ORDERLO + 1) * SUBFRAMES * ii], + &tmpLPCcoeffs_hi[(ORDERHI + 1)*SUBFRAMES * ii], + &tmpLPCindex_g[KLT_ORDER_GAIN * ii]); + } + + /* Entropy coding of quantization indices - LPC gain. */ + WebRtcIsac_EncHistMulti(ISACBitStr_obj, &tmpLPCindex_g[KLT_ORDER_GAIN * ii], + WebRtcIsac_kQKltCdfPtrGain, KLT_ORDER_GAIN); + + /* Quantization and loss-less coding. */ + status = WebRtcIsac_EncodeSpec(&tmp_fre[ii * FRAMESAMPLES_HALF], + &tmp_fim[ii * FRAMESAMPLES_HALF], + ISACSavedEnc_obj->AvgPitchGain[ii], + kIsacLowerBand, ISACBitStr_obj); + if (status < 0) { + return status; + } + } + /* Complete arithmetic coding. */ + return WebRtcIsac_EncTerminate(ISACBitStr_obj); +} + + +int WebRtcIsac_EncodeStoredDataUb( + const ISACUBSaveEncDataStruct* ISACSavedEnc_obj, + Bitstr* bitStream, + int32_t jitterInfo, + float scale, + enum ISACBandwidth bandwidth) { + int n; + int err; + double lpcGain[SUBFRAMES]; + int16_t realFFT[FRAMESAMPLES_HALF]; + int16_t imagFFT[FRAMESAMPLES_HALF]; + const uint16_t** shape_cdf; + int shape_len; + const int16_t kAveragePitchGain = 0.0; + enum ISACBand band; + /* Reset bitstream. */ + WebRtcIsac_ResetBitstream(bitStream); + + /* Encode jitter index. */ + WebRtcIsac_EncodeJitterInfo(jitterInfo, bitStream); + + err = WebRtcIsac_EncodeBandwidth(bandwidth, bitStream); + if (err < 0) { + return err; + } + + /* Encode LPC-shape. */ + if (bandwidth == isac12kHz) { + shape_cdf = WebRtcIsac_kLpcShapeCdfMatUb12; + shape_len = UB_LPC_ORDER * UB_LPC_VEC_PER_FRAME; + band = kIsacUpperBand12; + } else { + shape_cdf = WebRtcIsac_kLpcShapeCdfMatUb16; + shape_len = UB_LPC_ORDER * UB16_LPC_VEC_PER_FRAME; + band = kIsacUpperBand16; + } + WebRtcIsac_EncHistMulti(bitStream, ISACSavedEnc_obj->indexLPCShape, + shape_cdf, shape_len); + + if ((scale <= 0.0) || (scale >= 1.0)) { + /* We only consider scales between zero and one. */ + WebRtcIsac_EncHistMulti(bitStream, ISACSavedEnc_obj->lpcGainIndex, + WebRtcIsac_kLpcGainCdfMat, UB_LPC_GAIN_DIM); + if (bandwidth == isac16kHz) { + /* Store gain indices of the second half. */ + WebRtcIsac_EncHistMulti(bitStream, + &ISACSavedEnc_obj->lpcGainIndex[SUBFRAMES], + WebRtcIsac_kLpcGainCdfMat, UB_LPC_GAIN_DIM); + } + /* Store FFT coefficients. */ + err = WebRtcIsac_EncodeSpec(ISACSavedEnc_obj->realFFT, + ISACSavedEnc_obj->imagFFT, kAveragePitchGain, + band, bitStream); + } else { + /* Scale LPC gain and FFT coefficients. */ + for (n = 0; n < SUBFRAMES; n++) { + lpcGain[n] = scale * ISACSavedEnc_obj->lpcGain[n]; + } + /* Store LPC gains. */ + WebRtcIsac_StoreLpcGainUb(lpcGain, bitStream); + + if (bandwidth == isac16kHz) { + /* Scale and code the gains of the second half of the frame, if 16kHz. */ + for (n = 0; n < SUBFRAMES; n++) { + lpcGain[n] = scale * ISACSavedEnc_obj->lpcGain[n + SUBFRAMES]; + } + WebRtcIsac_StoreLpcGainUb(lpcGain, bitStream); + } + + for (n = 0; n < FRAMESAMPLES_HALF; n++) { + realFFT[n] = (int16_t)(scale * (float)ISACSavedEnc_obj->realFFT[n] + + 0.5f); + imagFFT[n] = (int16_t)(scale * (float)ISACSavedEnc_obj->imagFFT[n] + + 0.5f); + } + /* Store FFT coefficients. */ + err = WebRtcIsac_EncodeSpec(realFFT, imagFFT, kAveragePitchGain, + band, bitStream); + } + if (err < 0) { + /* Error happened while encoding FFT coefficients. */ + return err; + } + + /* Complete arithmetic coding. */ + return WebRtcIsac_EncTerminate(bitStream); +} + +int16_t WebRtcIsac_GetRedPayloadUb( + const ISACUBSaveEncDataStruct* ISACSavedEncObj, + Bitstr* bitStreamObj, + enum ISACBandwidth bandwidth) { + int n; + int16_t status; + int16_t realFFT[FRAMESAMPLES_HALF]; + int16_t imagFFT[FRAMESAMPLES_HALF]; + enum ISACBand band; + const int16_t kAveragePitchGain = 0.0; + /* Store bit-stream object. */ + memcpy(bitStreamObj, &ISACSavedEncObj->bitStreamObj, sizeof(Bitstr)); + + /* Scale FFT coefficients. */ + for (n = 0; n < FRAMESAMPLES_HALF; n++) { + realFFT[n] = (int16_t)((float)ISACSavedEncObj->realFFT[n] * + RCU_TRANSCODING_SCALE_UB + 0.5); + imagFFT[n] = (int16_t)((float)ISACSavedEncObj->imagFFT[n] * + RCU_TRANSCODING_SCALE_UB + 0.5); + } + + band = (bandwidth == isac12kHz) ? kIsacUpperBand12 : kIsacUpperBand16; + status = WebRtcIsac_EncodeSpec(realFFT, imagFFT, kAveragePitchGain, band, + bitStreamObj); + if (status < 0) { + return status; + } else { + /* Terminate entropy coding */ + return WebRtcIsac_EncTerminate(bitStreamObj); + } +} diff --git a/webrtc/modules/audio_coding/codecs/isac/main/source/fft.c b/webrtc/modules/audio_coding/codecs/isac/main/source/fft.c new file mode 100644 index 0000000..c854d8c --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/main/source/fft.c @@ -0,0 +1,943 @@ +/* + * Copyright(c)1995,97 Mark Olesen + * Queen's Univ at Kingston (Canada) + * + * Permission to use, copy, modify, and distribute this software for + * any purpose without fee is hereby granted, provided that this + * entire notice is included in all copies of any software which is + * or includes a copy or modification of this software and in all + * copies of the supporting documentation for such software. + * + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR + * IMPLIED WARRANTY. IN PARTICULAR, NEITHER THE AUTHOR NOR QUEEN'S + * UNIVERSITY AT KINGSTON MAKES ANY REPRESENTATION OR WARRANTY OF ANY + * KIND CONCERNING THE MERCHANTABILITY OF THIS SOFTWARE OR ITS + * FITNESS FOR ANY PARTICULAR PURPOSE. + * + * All of which is to say that you can do what you like with this + * source code provided you don't try to sell it as your own and you + * include an unaltered copy of this message (including the + * copyright). + * + * It is also implicitly understood that bug fixes and improvements + * should make their way back to the general Internet community so + * that everyone benefits. + * + * Changes: + * Trivial type modifications by the WebRTC authors. + */ + + +/* + * File: + * WebRtcIsac_Fftn.c + * + * Public: + * WebRtcIsac_Fftn / fftnf (); + * + * Private: + * WebRtcIsac_Fftradix / fftradixf (); + * + * Descript: + * multivariate complex Fourier transform, computed in place + * using mixed-radix Fast Fourier Transform algorithm. + * + * Fortran code by: + * RC Singleton, Stanford Research Institute, Sept. 1968 + * + * translated by f2c (version 19950721). + * + * int WebRtcIsac_Fftn (int ndim, const int dims[], REAL Re[], REAL Im[], + * int iSign, double scaling); + * + * NDIM = the total number dimensions + * DIMS = a vector of array sizes + * if NDIM is zero then DIMS must be zero-terminated + * + * RE and IM hold the real and imaginary components of the data, and return + * the resulting real and imaginary Fourier coefficients. Multidimensional + * data *must* be allocated contiguously. There is no limit on the number + * of dimensions. + * + * ISIGN = the sign of the complex exponential (ie, forward or inverse FFT) + * the magnitude of ISIGN (normally 1) is used to determine the + * correct indexing increment (see below). + * + * SCALING = normalizing constant by which the final result is *divided* + * if SCALING == -1, normalize by total dimension of the transform + * if SCALING < -1, normalize by the square-root of the total dimension + * + * example: + * tri-variate transform with Re[n1][n2][n3], Im[n1][n2][n3] + * + * int dims[3] = {n1,n2,n3} + * WebRtcIsac_Fftn (3, dims, Re, Im, 1, scaling); + * + *-----------------------------------------------------------------------* + * int WebRtcIsac_Fftradix (REAL Re[], REAL Im[], size_t nTotal, size_t nPass, + * size_t nSpan, int iSign, size_t max_factors, + * size_t max_perm); + * + * RE, IM - see above documentation + * + * Although there is no limit on the number of dimensions, WebRtcIsac_Fftradix() must + * be called once for each dimension, but the calls may be in any order. + * + * NTOTAL = the total number of complex data values + * NPASS = the dimension of the current variable + * NSPAN/NPASS = the spacing of consecutive data values while indexing the + * current variable + * ISIGN - see above documentation + * + * example: + * tri-variate transform with Re[n1][n2][n3], Im[n1][n2][n3] + * + * WebRtcIsac_Fftradix (Re, Im, n1*n2*n3, n1, n1, 1, maxf, maxp); + * WebRtcIsac_Fftradix (Re, Im, n1*n2*n3, n2, n1*n2, 1, maxf, maxp); + * WebRtcIsac_Fftradix (Re, Im, n1*n2*n3, n3, n1*n2*n3, 1, maxf, maxp); + * + * single-variate transform, + * NTOTAL = N = NSPAN = (number of complex data values), + * + * WebRtcIsac_Fftradix (Re, Im, n, n, n, 1, maxf, maxp); + * + * The data can also be stored in a single array with alternating real and + * imaginary parts, the magnitude of ISIGN is changed to 2 to give correct + * indexing increment, and data [0] and data [1] used to pass the initial + * addresses for the sequences of real and imaginary values, + * + * example: + * REAL data [2*NTOTAL]; + * WebRtcIsac_Fftradix ( &data[0], &data[1], NTOTAL, nPass, nSpan, 2, maxf, maxp); + * + * for temporary allocation: + * + * MAX_FACTORS >= the maximum prime factor of NPASS + * MAX_PERM >= the number of prime factors of NPASS. In addition, + * if the square-free portion K of NPASS has two or more prime + * factors, then MAX_PERM >= (K-1) + * + * storage in FACTOR for a maximum of 15 prime factors of NPASS. if NPASS + * has more than one square-free factor, the product of the square-free + * factors must be <= 210 array storage for maximum prime factor of 23 the + * following two constants should agree with the array dimensions. + * + *----------------------------------------------------------------------*/ +#include "fft.h" + +#include +#include + + + +/* double precision routine */ +static int +WebRtcIsac_Fftradix (double Re[], double Im[], + size_t nTotal, size_t nPass, size_t nSpan, int isign, + int max_factors, unsigned int max_perm, + FFTstr *fftstate); + + + +#ifndef M_PI +# define M_PI 3.14159265358979323846264338327950288 +#endif + +#ifndef SIN60 +# define SIN60 0.86602540378443865 /* sin(60 deg) */ +# define COS72 0.30901699437494742 /* cos(72 deg) */ +# define SIN72 0.95105651629515357 /* sin(72 deg) */ +#endif + +# define REAL double +# define FFTN WebRtcIsac_Fftn +# define FFTNS "fftn" +# define FFTRADIX WebRtcIsac_Fftradix +# define FFTRADIXS "fftradix" + + +int WebRtcIsac_Fftns(unsigned int ndim, const int dims[], + double Re[], + double Im[], + int iSign, + double scaling, + FFTstr *fftstate) +{ + + size_t nSpan, nPass, nTotal; + unsigned int i; + int ret, max_factors, max_perm; + + /* + * tally the number of elements in the data array + * and determine the number of dimensions + */ + nTotal = 1; + if (ndim && dims [0]) + { + for (i = 0; i < ndim; i++) + { + if (dims [i] <= 0) + { + return -1; + } + nTotal *= dims [i]; + } + } + else + { + ndim = 0; + for (i = 0; dims [i]; i++) + { + if (dims [i] <= 0) + { + return -1; + } + nTotal *= dims [i]; + ndim++; + } + } + + /* determine maximum number of factors and permuations */ +#if 1 + /* + * follow John Beale's example, just use the largest dimension and don't + * worry about excess allocation. May be someone else will do it? + */ + max_factors = max_perm = 1; + for (i = 0; i < ndim; i++) + { + nSpan = dims [i]; + if ((int)nSpan > max_factors) + { + max_factors = (int)nSpan; + } + if ((int)nSpan > max_perm) + { + max_perm = (int)nSpan; + } + } +#else + /* use the constants used in the original Fortran code */ + max_factors = 23; + max_perm = 209; +#endif + /* loop over the dimensions: */ + nPass = 1; + for (i = 0; i < ndim; i++) + { + nSpan = dims [i]; + nPass *= nSpan; + ret = FFTRADIX (Re, Im, nTotal, nSpan, nPass, iSign, + max_factors, max_perm, fftstate); + /* exit, clean-up already done */ + if (ret) + return ret; + } + + /* Divide through by the normalizing constant: */ + if (scaling && scaling != 1.0) + { + if (iSign < 0) iSign = -iSign; + if (scaling < 0.0) + { + scaling = (double)nTotal; + if (scaling < -1.0) + scaling = sqrt (scaling); + } + scaling = 1.0 / scaling; /* multiply is often faster */ + for (i = 0; i < nTotal; i += iSign) + { + Re [i] *= scaling; + Im [i] *= scaling; + } + } + return 0; +} + +/* + * singleton's mixed radix routine + * + * could move allocation out to WebRtcIsac_Fftn(), but leave it here so that it's + * possible to make this a standalone function + */ + +static int FFTRADIX (REAL Re[], + REAL Im[], + size_t nTotal, + size_t nPass, + size_t nSpan, + int iSign, + int max_factors, + unsigned int max_perm, + FFTstr *fftstate) +{ + int ii, mfactor, kspan, ispan, inc; + int j, jc, jf, jj, k, k1, k2, k3, k4, kk, kt, nn, ns, nt; + + + REAL radf; + REAL c1, c2, c3, cd, aa, aj, ak, ajm, ajp, akm, akp; + REAL s1, s2, s3, sd, bb, bj, bk, bjm, bjp, bkm, bkp; + + REAL *Rtmp = NULL; /* temp space for real part*/ + REAL *Itmp = NULL; /* temp space for imaginary part */ + REAL *Cos = NULL; /* Cosine values */ + REAL *Sin = NULL; /* Sine values */ + + REAL s60 = SIN60; /* sin(60 deg) */ + REAL c72 = COS72; /* cos(72 deg) */ + REAL s72 = SIN72; /* sin(72 deg) */ + REAL pi2 = M_PI; /* use PI first, 2 PI later */ + + + fftstate->SpaceAlloced = 0; + fftstate->MaxPermAlloced = 0; + + + // initialize to avoid warnings + k3 = c2 = c3 = s2 = s3 = 0.0; + + if (nPass < 2) + return 0; + + /* allocate storage */ + if (fftstate->SpaceAlloced < max_factors * sizeof (REAL)) + { +#ifdef SUN_BROKEN_REALLOC + if (!fftstate->SpaceAlloced) /* first time */ + { + fftstate->SpaceAlloced = max_factors * sizeof (REAL); + } + else + { +#endif + fftstate->SpaceAlloced = max_factors * sizeof (REAL); +#ifdef SUN_BROKEN_REALLOC + } +#endif + } + else + { + /* allow full use of alloc'd space */ + max_factors = fftstate->SpaceAlloced / sizeof (REAL); + } + if (fftstate->MaxPermAlloced < max_perm) + { +#ifdef SUN_BROKEN_REALLOC + if (!fftstate->MaxPermAlloced) /* first time */ + else +#endif + fftstate->MaxPermAlloced = max_perm; + } + else + { + /* allow full use of alloc'd space */ + max_perm = fftstate->MaxPermAlloced; + } + + /* assign pointers */ + Rtmp = (REAL *) fftstate->Tmp0; + Itmp = (REAL *) fftstate->Tmp1; + Cos = (REAL *) fftstate->Tmp2; + Sin = (REAL *) fftstate->Tmp3; + + /* + * Function Body + */ + inc = iSign; + if (iSign < 0) { + s72 = -s72; + s60 = -s60; + pi2 = -pi2; + inc = -inc; /* absolute value */ + } + + /* adjust for strange increments */ + nt = inc * (int)nTotal; + ns = inc * (int)nSpan; + kspan = ns; + + nn = nt - inc; + jc = ns / (int)nPass; + radf = pi2 * (double) jc; + pi2 *= 2.0; /* use 2 PI from here on */ + + ii = 0; + jf = 0; + /* determine the factors of n */ + mfactor = 0; + k = (int)nPass; + while (k % 16 == 0) { + mfactor++; + fftstate->factor [mfactor - 1] = 4; + k /= 16; + } + j = 3; + jj = 9; + do { + while (k % jj == 0) { + mfactor++; + fftstate->factor [mfactor - 1] = j; + k /= jj; + } + j += 2; + jj = j * j; + } while (jj <= k); + if (k <= 4) { + kt = mfactor; + fftstate->factor [mfactor] = k; + if (k != 1) + mfactor++; + } else { + if (k - (k / 4 << 2) == 0) { + mfactor++; + fftstate->factor [mfactor - 1] = 2; + k /= 4; + } + kt = mfactor; + j = 2; + do { + if (k % j == 0) { + mfactor++; + fftstate->factor [mfactor - 1] = j; + k /= j; + } + j = ((j + 1) / 2 << 1) + 1; + } while (j <= k); + } + if (kt) { + j = kt; + do { + mfactor++; + fftstate->factor [mfactor - 1] = fftstate->factor [j - 1]; + j--; + } while (j); + } + + /* test that mfactors is in range */ + if (mfactor > NFACTOR) + { + return -1; + } + + /* compute fourier transform */ + for (;;) { + sd = radf / (double) kspan; + cd = sin(sd); + cd = 2.0 * cd * cd; + sd = sin(sd + sd); + kk = 0; + ii++; + + switch (fftstate->factor [ii - 1]) { + case 2: + /* transform for factor of 2 (including rotation factor) */ + kspan /= 2; + k1 = kspan + 2; + do { + do { + k2 = kk + kspan; + ak = Re [k2]; + bk = Im [k2]; + Re [k2] = Re [kk] - ak; + Im [k2] = Im [kk] - bk; + Re [kk] += ak; + Im [kk] += bk; + kk = k2 + kspan; + } while (kk < nn); + kk -= nn; + } while (kk < jc); + if (kk >= kspan) + goto Permute_Results_Label; /* exit infinite loop */ + do { + c1 = 1.0 - cd; + s1 = sd; + do { + do { + do { + k2 = kk + kspan; + ak = Re [kk] - Re [k2]; + bk = Im [kk] - Im [k2]; + Re [kk] += Re [k2]; + Im [kk] += Im [k2]; + Re [k2] = c1 * ak - s1 * bk; + Im [k2] = s1 * ak + c1 * bk; + kk = k2 + kspan; + } while (kk < (nt-1)); + k2 = kk - nt; + c1 = -c1; + kk = k1 - k2; + } while (kk > k2); + ak = c1 - (cd * c1 + sd * s1); + s1 = sd * c1 - cd * s1 + s1; + c1 = 2.0 - (ak * ak + s1 * s1); + s1 *= c1; + c1 *= ak; + kk += jc; + } while (kk < k2); + k1 += inc + inc; + kk = (k1 - kspan + 1) / 2 + jc - 1; + } while (kk < (jc + jc)); + break; + + case 4: /* transform for factor of 4 */ + ispan = kspan; + kspan /= 4; + + do { + c1 = 1.0; + s1 = 0.0; + do { + do { + k1 = kk + kspan; + k2 = k1 + kspan; + k3 = k2 + kspan; + akp = Re [kk] + Re [k2]; + akm = Re [kk] - Re [k2]; + ajp = Re [k1] + Re [k3]; + ajm = Re [k1] - Re [k3]; + bkp = Im [kk] + Im [k2]; + bkm = Im [kk] - Im [k2]; + bjp = Im [k1] + Im [k3]; + bjm = Im [k1] - Im [k3]; + Re [kk] = akp + ajp; + Im [kk] = bkp + bjp; + ajp = akp - ajp; + bjp = bkp - bjp; + if (iSign < 0) { + akp = akm + bjm; + bkp = bkm - ajm; + akm -= bjm; + bkm += ajm; + } else { + akp = akm - bjm; + bkp = bkm + ajm; + akm += bjm; + bkm -= ajm; + } + /* avoid useless multiplies */ + if (s1 == 0.0) { + Re [k1] = akp; + Re [k2] = ajp; + Re [k3] = akm; + Im [k1] = bkp; + Im [k2] = bjp; + Im [k3] = bkm; + } else { + Re [k1] = akp * c1 - bkp * s1; + Re [k2] = ajp * c2 - bjp * s2; + Re [k3] = akm * c3 - bkm * s3; + Im [k1] = akp * s1 + bkp * c1; + Im [k2] = ajp * s2 + bjp * c2; + Im [k3] = akm * s3 + bkm * c3; + } + kk = k3 + kspan; + } while (kk < nt); + + c2 = c1 - (cd * c1 + sd * s1); + s1 = sd * c1 - cd * s1 + s1; + c1 = 2.0 - (c2 * c2 + s1 * s1); + s1 *= c1; + c1 *= c2; + /* values of c2, c3, s2, s3 that will get used next time */ + c2 = c1 * c1 - s1 * s1; + s2 = 2.0 * c1 * s1; + c3 = c2 * c1 - s2 * s1; + s3 = c2 * s1 + s2 * c1; + kk = kk - nt + jc; + } while (kk < kspan); + kk = kk - kspan + inc; + } while (kk < jc); + if (kspan == jc) + goto Permute_Results_Label; /* exit infinite loop */ + break; + + default: + /* transform for odd factors */ +#ifdef FFT_RADIX4 + return -1; + break; +#else /* FFT_RADIX4 */ + k = fftstate->factor [ii - 1]; + ispan = kspan; + kspan /= k; + + switch (k) { + case 3: /* transform for factor of 3 (optional code) */ + do { + do { + k1 = kk + kspan; + k2 = k1 + kspan; + ak = Re [kk]; + bk = Im [kk]; + aj = Re [k1] + Re [k2]; + bj = Im [k1] + Im [k2]; + Re [kk] = ak + aj; + Im [kk] = bk + bj; + ak -= 0.5 * aj; + bk -= 0.5 * bj; + aj = (Re [k1] - Re [k2]) * s60; + bj = (Im [k1] - Im [k2]) * s60; + Re [k1] = ak - bj; + Re [k2] = ak + bj; + Im [k1] = bk + aj; + Im [k2] = bk - aj; + kk = k2 + kspan; + } while (kk < (nn - 1)); + kk -= nn; + } while (kk < kspan); + break; + + case 5: /* transform for factor of 5 (optional code) */ + c2 = c72 * c72 - s72 * s72; + s2 = 2.0 * c72 * s72; + do { + do { + k1 = kk + kspan; + k2 = k1 + kspan; + k3 = k2 + kspan; + k4 = k3 + kspan; + akp = Re [k1] + Re [k4]; + akm = Re [k1] - Re [k4]; + bkp = Im [k1] + Im [k4]; + bkm = Im [k1] - Im [k4]; + ajp = Re [k2] + Re [k3]; + ajm = Re [k2] - Re [k3]; + bjp = Im [k2] + Im [k3]; + bjm = Im [k2] - Im [k3]; + aa = Re [kk]; + bb = Im [kk]; + Re [kk] = aa + akp + ajp; + Im [kk] = bb + bkp + bjp; + ak = akp * c72 + ajp * c2 + aa; + bk = bkp * c72 + bjp * c2 + bb; + aj = akm * s72 + ajm * s2; + bj = bkm * s72 + bjm * s2; + Re [k1] = ak - bj; + Re [k4] = ak + bj; + Im [k1] = bk + aj; + Im [k4] = bk - aj; + ak = akp * c2 + ajp * c72 + aa; + bk = bkp * c2 + bjp * c72 + bb; + aj = akm * s2 - ajm * s72; + bj = bkm * s2 - bjm * s72; + Re [k2] = ak - bj; + Re [k3] = ak + bj; + Im [k2] = bk + aj; + Im [k3] = bk - aj; + kk = k4 + kspan; + } while (kk < (nn-1)); + kk -= nn; + } while (kk < kspan); + break; + + default: + if (k != jf) { + jf = k; + s1 = pi2 / (double) k; + c1 = cos(s1); + s1 = sin(s1); + if (jf > max_factors){ + return -1; + } + Cos [jf - 1] = 1.0; + Sin [jf - 1] = 0.0; + j = 1; + do { + Cos [j - 1] = Cos [k - 1] * c1 + Sin [k - 1] * s1; + Sin [j - 1] = Cos [k - 1] * s1 - Sin [k - 1] * c1; + k--; + Cos [k - 1] = Cos [j - 1]; + Sin [k - 1] = -Sin [j - 1]; + j++; + } while (j < k); + } + do { + do { + k1 = kk; + k2 = kk + ispan; + ak = aa = Re [kk]; + bk = bb = Im [kk]; + j = 1; + k1 += kspan; + do { + k2 -= kspan; + j++; + Rtmp [j - 1] = Re [k1] + Re [k2]; + ak += Rtmp [j - 1]; + Itmp [j - 1] = Im [k1] + Im [k2]; + bk += Itmp [j - 1]; + j++; + Rtmp [j - 1] = Re [k1] - Re [k2]; + Itmp [j - 1] = Im [k1] - Im [k2]; + k1 += kspan; + } while (k1 < k2); + Re [kk] = ak; + Im [kk] = bk; + k1 = kk; + k2 = kk + ispan; + j = 1; + do { + k1 += kspan; + k2 -= kspan; + jj = j; + ak = aa; + bk = bb; + aj = 0.0; + bj = 0.0; + k = 1; + do { + k++; + ak += Rtmp [k - 1] * Cos [jj - 1]; + bk += Itmp [k - 1] * Cos [jj - 1]; + k++; + aj += Rtmp [k - 1] * Sin [jj - 1]; + bj += Itmp [k - 1] * Sin [jj - 1]; + jj += j; + if (jj > jf) { + jj -= jf; + } + } while (k < jf); + k = jf - j; + Re [k1] = ak - bj; + Im [k1] = bk + aj; + Re [k2] = ak + bj; + Im [k2] = bk - aj; + j++; + } while (j < k); + kk += ispan; + } while (kk < nn); + kk -= nn; + } while (kk < kspan); + break; + } + + /* multiply by rotation factor (except for factors of 2 and 4) */ + if (ii == mfactor) + goto Permute_Results_Label; /* exit infinite loop */ + kk = jc; + do { + c2 = 1.0 - cd; + s1 = sd; + do { + c1 = c2; + s2 = s1; + kk += kspan; + do { + do { + ak = Re [kk]; + Re [kk] = c2 * ak - s2 * Im [kk]; + Im [kk] = s2 * ak + c2 * Im [kk]; + kk += ispan; + } while (kk < nt); + ak = s1 * s2; + s2 = s1 * c2 + c1 * s2; + c2 = c1 * c2 - ak; + kk = kk - nt + kspan; + } while (kk < ispan); + c2 = c1 - (cd * c1 + sd * s1); + s1 += sd * c1 - cd * s1; + c1 = 2.0 - (c2 * c2 + s1 * s1); + s1 *= c1; + c2 *= c1; + kk = kk - ispan + jc; + } while (kk < kspan); + kk = kk - kspan + jc + inc; + } while (kk < (jc + jc)); + break; +#endif /* FFT_RADIX4 */ + } + } + + /* permute the results to normal order---done in two stages */ + /* permutation for square factors of n */ +Permute_Results_Label: + fftstate->Perm [0] = ns; + if (kt) { + k = kt + kt + 1; + if (mfactor < k) + k--; + j = 1; + fftstate->Perm [k] = jc; + do { + fftstate->Perm [j] = fftstate->Perm [j - 1] / fftstate->factor [j - 1]; + fftstate->Perm [k - 1] = fftstate->Perm [k] * fftstate->factor [j - 1]; + j++; + k--; + } while (j < k); + k3 = fftstate->Perm [k]; + kspan = fftstate->Perm [1]; + kk = jc; + k2 = kspan; + j = 1; + if (nPass != nTotal) { + /* permutation for multivariate transform */ + Permute_Multi_Label: + do { + do { + k = kk + jc; + do { + /* swap Re [kk] <> Re [k2], Im [kk] <> Im [k2] */ + ak = Re [kk]; Re [kk] = Re [k2]; Re [k2] = ak; + bk = Im [kk]; Im [kk] = Im [k2]; Im [k2] = bk; + kk += inc; + k2 += inc; + } while (kk < (k-1)); + kk += ns - jc; + k2 += ns - jc; + } while (kk < (nt-1)); + k2 = k2 - nt + kspan; + kk = kk - nt + jc; + } while (k2 < (ns-1)); + do { + do { + k2 -= fftstate->Perm [j - 1]; + j++; + k2 = fftstate->Perm [j] + k2; + } while (k2 > fftstate->Perm [j - 1]); + j = 1; + do { + if (kk < (k2-1)) + goto Permute_Multi_Label; + kk += jc; + k2 += kspan; + } while (k2 < (ns-1)); + } while (kk < (ns-1)); + } else { + /* permutation for single-variate transform (optional code) */ + Permute_Single_Label: + do { + /* swap Re [kk] <> Re [k2], Im [kk] <> Im [k2] */ + ak = Re [kk]; Re [kk] = Re [k2]; Re [k2] = ak; + bk = Im [kk]; Im [kk] = Im [k2]; Im [k2] = bk; + kk += inc; + k2 += kspan; + } while (k2 < (ns-1)); + do { + do { + k2 -= fftstate->Perm [j - 1]; + j++; + k2 = fftstate->Perm [j] + k2; + } while (k2 >= fftstate->Perm [j - 1]); + j = 1; + do { + if (kk < k2) + goto Permute_Single_Label; + kk += inc; + k2 += kspan; + } while (k2 < (ns-1)); + } while (kk < (ns-1)); + } + jc = k3; + } + + if ((kt << 1) + 1 >= mfactor) + return 0; + ispan = fftstate->Perm [kt]; + /* permutation for square-free factors of n */ + j = mfactor - kt; + fftstate->factor [j] = 1; + do { + fftstate->factor [j - 1] *= fftstate->factor [j]; + j--; + } while (j != kt); + kt++; + nn = fftstate->factor [kt - 1] - 1; + if (nn > (int) max_perm) { + return -1; + } + j = jj = 0; + for (;;) { + k = kt + 1; + k2 = fftstate->factor [kt - 1]; + kk = fftstate->factor [k - 1]; + j++; + if (j > nn) + break; /* exit infinite loop */ + jj += kk; + while (jj >= k2) { + jj -= k2; + k2 = kk; + k++; + kk = fftstate->factor [k - 1]; + jj += kk; + } + fftstate->Perm [j - 1] = jj; + } + /* determine the permutation cycles of length greater than 1 */ + j = 0; + for (;;) { + do { + j++; + kk = fftstate->Perm [j - 1]; + } while (kk < 0); + if (kk != j) { + do { + k = kk; + kk = fftstate->Perm [k - 1]; + fftstate->Perm [k - 1] = -kk; + } while (kk != j); + k3 = kk; + } else { + fftstate->Perm [j - 1] = -j; + if (j == nn) + break; /* exit infinite loop */ + } + } + max_factors *= inc; + /* reorder a and b, following the permutation cycles */ + for (;;) { + j = k3 + 1; + nt -= ispan; + ii = nt - inc + 1; + if (nt < 0) + break; /* exit infinite loop */ + do { + do { + j--; + } while (fftstate->Perm [j - 1] < 0); + jj = jc; + do { + kspan = jj; + if (jj > max_factors) { + kspan = max_factors; + } + jj -= kspan; + k = fftstate->Perm [j - 1]; + kk = jc * k + ii + jj; + k1 = kk + kspan - 1; + k2 = 0; + do { + k2++; + Rtmp [k2 - 1] = Re [k1]; + Itmp [k2 - 1] = Im [k1]; + k1 -= inc; + } while (k1 != (kk-1)); + do { + k1 = kk + kspan - 1; + k2 = k1 - jc * (k + fftstate->Perm [k - 1]); + k = -fftstate->Perm [k - 1]; + do { + Re [k1] = Re [k2]; + Im [k1] = Im [k2]; + k1 -= inc; + k2 -= inc; + } while (k1 != (kk-1)); + kk = k2 + 1; + } while (k != j); + k1 = kk + kspan - 1; + k2 = 0; + do { + k2++; + Re [k1] = Rtmp [k2 - 1]; + Im [k1] = Itmp [k2 - 1]; + k1 -= inc; + } while (k1 != (kk-1)); + } while (jj); + } while (j != 1); + } + return 0; /* exit point here */ +} +/* ---------------------- end-of-file (c source) ---------------------- */ + diff --git a/webrtc/modules/audio_coding/codecs/isac/main/source/fft.h b/webrtc/modules/audio_coding/codecs/isac/main/source/fft.h new file mode 100644 index 0000000..a42f57b --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/main/source/fft.h @@ -0,0 +1,45 @@ +/* + * Copyright (c) 2011 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +/*--------------------------------*-C-*---------------------------------* + * File: + * fftn.h + * ---------------------------------------------------------------------* + * Re[]: real value array + * Im[]: imaginary value array + * nTotal: total number of complex values + * nPass: number of elements involved in this pass of transform + * nSpan: nspan/nPass = number of bytes to increment pointer + * in Re[] and Im[] + * isign: exponent: +1 = forward -1 = reverse + * scaling: normalizing constant by which the final result is *divided* + * scaling == -1, normalize by total dimension of the transform + * scaling < -1, normalize by the square-root of the total dimension + * + * ---------------------------------------------------------------------- + * See the comments in the code for correct usage! + */ + +#ifndef WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_SOURCE_FFT_H_ +#define WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_SOURCE_FFT_H_ + + +#include "structs.h" + + +/* double precision routine */ + + +int WebRtcIsac_Fftns (unsigned int ndim, const int dims[], double Re[], double Im[], + int isign, double scaling, FFTstr *fftstate); + + + +#endif /* WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_SOURCE_FFT_H_ */ diff --git a/webrtc/modules/audio_coding/codecs/isac/main/source/isac.c b/webrtc/modules/audio_coding/codecs/isac/main/source/isac.c new file mode 100644 index 0000000..875e7ac --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/main/source/isac.c @@ -0,0 +1,2363 @@ +/* + * Copyright (c) 2012 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +/* + * isac.c + * + * This C file contains the functions for the ISAC API + * + */ + +#include "webrtc/modules/audio_coding/codecs/isac/main/include/isac.h" + +#include +#include +#include +#include +#include + +#include "webrtc/common_audio/signal_processing/include/signal_processing_library.h" +#include "webrtc/modules/audio_coding/codecs/isac/main/source/bandwidth_estimator.h" +#include "webrtc/modules/audio_coding/codecs/isac/main/source/codec.h" +#include "webrtc/modules/audio_coding/codecs/isac/main/source/crc.h" +#include "webrtc/modules/audio_coding/codecs/isac/main/source/entropy_coding.h" +#include "webrtc/modules/audio_coding/codecs/isac/main/source/lpc_shape_swb16_tables.h" +#include "webrtc/modules/audio_coding/codecs/isac/main/source/os_specific_inline.h" +#include "webrtc/modules/audio_coding/codecs/isac/main/source/structs.h" + +#define BIT_MASK_DEC_INIT 0x0001 +#define BIT_MASK_ENC_INIT 0x0002 + +#define LEN_CHECK_SUM_WORD8 4 +#define MAX_NUM_LAYERS 10 + + +/**************************************************************************** + * UpdatePayloadSizeLimit(...) + * + * Call this function to update the limit on the payload size. The limit on + * payload size might change i) if a user ''directly changes the limit by + * calling xxx_setMaxPayloadSize() or xxx_setMaxRate(), or ii) indirectly + * when bandwidth is changing. The latter might be the result of bandwidth + * adaptation, or direct change of the bottleneck in instantaneous mode. + * + * This function takes the current overall limit on payload, and translates it + * to the limits on lower and upper-band. If the codec is in wideband mode, + * then the overall limit and the limit on the lower-band is the same. + * Otherwise, a fraction of the limit should be allocated to lower-band + * leaving some room for the upper-band bit-stream. That is why an update + * of limit is required every time that the bandwidth is changing. + * + */ +static void UpdatePayloadSizeLimit(ISACMainStruct* instISAC) { + int16_t lim30MsPayloadBytes = WEBRTC_SPL_MIN( + (instISAC->maxPayloadSizeBytes), + (instISAC->maxRateBytesPer30Ms)); + int16_t lim60MsPayloadBytes = WEBRTC_SPL_MIN( + (instISAC->maxPayloadSizeBytes), + (instISAC->maxRateBytesPer30Ms << 1)); + + /* The only time that iSAC will have 60 ms + * frame-size is when operating in wideband, so + * there is no upper-band bit-stream. */ + + if (instISAC->bandwidthKHz == isac8kHz) { + /* At 8 kHz there is no upper-band bit-stream, + * therefore, the lower-band limit is the overall limit. */ + instISAC->instLB.ISACencLB_obj.payloadLimitBytes60 = + lim60MsPayloadBytes; + instISAC->instLB.ISACencLB_obj.payloadLimitBytes30 = + lim30MsPayloadBytes; + } else { + /* When in super-wideband, we only have 30 ms frames. + * Do a rate allocation for the given limit. */ + if (lim30MsPayloadBytes > 250) { + /* 4/5 to lower-band the rest for upper-band. */ + instISAC->instLB.ISACencLB_obj.payloadLimitBytes30 = + (lim30MsPayloadBytes << 2) / 5; + } else if (lim30MsPayloadBytes > 200) { + /* For the interval of 200 to 250 the share of + * upper-band linearly grows from 20 to 50. */ + instISAC->instLB.ISACencLB_obj.payloadLimitBytes30 = + (lim30MsPayloadBytes << 1) / 5 + 100; + } else { + /* Allocate only 20 for upper-band. */ + instISAC->instLB.ISACencLB_obj.payloadLimitBytes30 = + lim30MsPayloadBytes - 20; + } + instISAC->instUB.ISACencUB_obj.maxPayloadSizeBytes = + lim30MsPayloadBytes; + } +} + + +/**************************************************************************** + * UpdateBottleneck(...) + * + * This function updates the bottleneck only if the codec is operating in + * channel-adaptive mode. Furthermore, as the update of bottleneck might + * result in an update of bandwidth, therefore, the bottlenech should be + * updated just right before the first 10ms of a frame is pushed into encoder. + * + */ +static void UpdateBottleneck(ISACMainStruct* instISAC) { + /* Read the bottleneck from bandwidth estimator for the + * first 10 ms audio. This way, if there is a change + * in bandwidth, upper and lower-band will be in sync. */ + if ((instISAC->codingMode == 0) && + (instISAC->instLB.ISACencLB_obj.buffer_index == 0) && + (instISAC->instLB.ISACencLB_obj.frame_nb == 0)) { + int32_t bottleneck = + WebRtcIsac_GetUplinkBandwidth(&instISAC->bwestimator_obj); + + /* Adding hysteresis when increasing signal bandwidth. */ + if ((instISAC->bandwidthKHz == isac8kHz) + && (bottleneck > 37000) + && (bottleneck < 41000)) { + bottleneck = 37000; + } + + /* Switching from 12 kHz to 16 kHz is not allowed at this revision. + * If we let this happen, we have to take care of buffer_index and + * the last LPC vector. */ + if ((instISAC->bandwidthKHz != isac16kHz) && + (bottleneck > 46000)) { + bottleneck = 46000; + } + + /* We might need a rate allocation. */ + if (instISAC->encoderSamplingRateKHz == kIsacWideband) { + /* Wideband is the only choice we have here. */ + instISAC->instLB.ISACencLB_obj.bottleneck = + (bottleneck > 32000) ? 32000 : bottleneck; + instISAC->bandwidthKHz = isac8kHz; + } else { + /* Do the rate-allocation and get the new bandwidth. */ + enum ISACBandwidth bandwidth; + WebRtcIsac_RateAllocation(bottleneck, + &(instISAC->instLB.ISACencLB_obj.bottleneck), + &(instISAC->instUB.ISACencUB_obj.bottleneck), + &bandwidth); + if (bandwidth != isac8kHz) { + instISAC->instLB.ISACencLB_obj.new_framelength = 480; + } + if (bandwidth != instISAC->bandwidthKHz) { + /* Bandwidth is changing. */ + instISAC->bandwidthKHz = bandwidth; + UpdatePayloadSizeLimit(instISAC); + if (bandwidth == isac12kHz) { + instISAC->instLB.ISACencLB_obj.buffer_index = 0; + } + /* Currently we don't let the bandwidth to switch to 16 kHz + * if in adaptive mode. If we let this happen, we have to take + * care of buffer_index and the last LPC vector. */ + } + } + } +} + + +/**************************************************************************** + * GetSendBandwidthInfo(...) + * + * This is called to get the bandwidth info. This info is the bandwidth and + * the jitter of 'there-to-here' channel, estimated 'here.' These info + * is signaled in an in-band fashion to the other side. + * + * The call to the bandwidth estimator triggers a recursive averaging which + * has to be synchronized between encoder & decoder, therefore, the call to + * BWE should be once per packet. As the BWE info is inserted into bit-stream + * We need a valid info right before the encodeLB function is going to + * generate a bit-stream. That is when lower-band buffer has already 20ms + * of audio, and the 3rd block of 10ms is going to be injected into encoder. + * + * Inputs: + * - instISAC : iSAC instance. + * + * Outputs: + * - bandwidthIndex : an index which has to be encoded in + * lower-band bit-stream, indicating the + * bandwidth of there-to-here channel. + * - jitterInfo : this indicates if the jitter is high + * or low and it is encoded in upper-band + * bit-stream. + * + */ +static void GetSendBandwidthInfo(ISACMainStruct* instISAC, + int16_t* bandwidthIndex, + int16_t* jitterInfo) { + if ((instISAC->instLB.ISACencLB_obj.buffer_index == + (FRAMESAMPLES_10ms << 1)) && + (instISAC->instLB.ISACencLB_obj.frame_nb == 0)) { + /* Bandwidth estimation and coding. */ + WebRtcIsac_GetDownlinkBwJitIndexImpl(&(instISAC->bwestimator_obj), + bandwidthIndex, jitterInfo, + instISAC->decoderSamplingRateKHz); + } +} + + +/**************************************************************************** + * WebRtcIsac_AssignSize(...) + * + * This function returns the size of the ISAC instance, so that the instance + * can be created out side iSAC. + * + * Output: + * - sizeinbytes : number of bytes needed to allocate for the + * instance. + * + * Return value : 0 - Ok + * -1 - Error + */ +int16_t WebRtcIsac_AssignSize(int* sizeInBytes) { + *sizeInBytes = sizeof(ISACMainStruct) * 2 / sizeof(int16_t); + return 0; +} + + +/**************************************************************************** + * WebRtcIsac_Assign(...) + * + * This function assigns the memory already created to the ISAC instance. + * + * Input: + * - ISAC_main_inst : address of the pointer to the coder instance. + * - instISAC_Addr : the already allocated memory, where we put the + * iSAC structure. + * + * Return value : 0 - Ok + * -1 - Error + */ +int16_t WebRtcIsac_Assign(ISACStruct** ISAC_main_inst, + void* instISAC_Addr) { + if (instISAC_Addr != NULL) { + ISACMainStruct* instISAC = (ISACMainStruct*)instISAC_Addr; + instISAC->errorCode = 0; + instISAC->initFlag = 0; + + /* Assign the address. */ + *ISAC_main_inst = (ISACStruct*)instISAC_Addr; + + /* Default is wideband. */ + instISAC->encoderSamplingRateKHz = kIsacWideband; + instISAC->decoderSamplingRateKHz = kIsacWideband; + instISAC->bandwidthKHz = isac8kHz; + instISAC->in_sample_rate_hz = 16000; + + WebRtcIsac_InitTransform(&instISAC->transform_tables); + return 0; + } else { + return -1; + } +} + + +/**************************************************************************** + * WebRtcIsac_Create(...) + * + * This function creates an ISAC instance, which will contain the state + * information for one coding/decoding channel. + * + * Input: + * - ISAC_main_inst : address of the pointer to the coder instance. + * + * Return value : 0 - Ok + * -1 - Error + */ +int16_t WebRtcIsac_Create(ISACStruct** ISAC_main_inst) { + ISACMainStruct* instISAC; + + if (ISAC_main_inst != NULL) { + instISAC = (ISACMainStruct*)malloc(sizeof(ISACMainStruct)); + *ISAC_main_inst = (ISACStruct*)instISAC; + if (*ISAC_main_inst != NULL) { + instISAC->errorCode = 0; + instISAC->initFlag = 0; + /* Default is wideband. */ + instISAC->bandwidthKHz = isac8kHz; + instISAC->encoderSamplingRateKHz = kIsacWideband; + instISAC->decoderSamplingRateKHz = kIsacWideband; + instISAC->in_sample_rate_hz = 16000; + + WebRtcIsac_InitTransform(&instISAC->transform_tables); + return 0; + } else { + return -1; + } + } else { + return -1; + } +} + + +/**************************************************************************** + * WebRtcIsac_Free(...) + * + * This function frees the ISAC instance created at the beginning. + * + * Input: + * - ISAC_main_inst : a ISAC instance. + * + * Return value : 0 - Ok + * -1 - Error + */ +int16_t WebRtcIsac_Free(ISACStruct* ISAC_main_inst) { + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + free(instISAC); + return 0; +} + + +/**************************************************************************** + * EncoderInitLb(...) - internal function for initialization of + * Lower Band + * EncoderInitUb(...) - internal function for initialization of + * Upper Band + * WebRtcIsac_EncoderInit(...) - API function + * + * This function initializes a ISAC instance prior to the encoder calls. + * + * Input: + * - ISAC_main_inst : ISAC instance. + * - CodingMode : 0 -> Bit rate and frame length are automatically + * adjusted to available bandwidth on + * transmission channel, applicable just to + * wideband mode. + * 1 -> User sets a frame length and a target bit + * rate which is taken as the maximum + * short-term average bit rate. + * + * Return value : 0 - Ok + * -1 - Error + */ +static int16_t EncoderInitLb(ISACLBStruct* instLB, + int16_t codingMode, + enum IsacSamplingRate sampRate) { + int16_t statusInit = 0; + int k; + + /* Init stream vector to zero */ + for (k = 0; k < STREAM_SIZE_MAX_60; k++) { + instLB->ISACencLB_obj.bitstr_obj.stream[k] = 0; + } + + if ((codingMode == 1) || (sampRate == kIsacSuperWideband)) { + /* 30 ms frame-size if either in super-wideband or + * instantaneous mode (I-mode). */ + instLB->ISACencLB_obj.new_framelength = 480; + } else { + instLB->ISACencLB_obj.new_framelength = INITIAL_FRAMESAMPLES; + } + + WebRtcIsac_InitMasking(&instLB->ISACencLB_obj.maskfiltstr_obj); + WebRtcIsac_InitPreFilterbank(&instLB->ISACencLB_obj.prefiltbankstr_obj); + WebRtcIsac_InitPitchFilter(&instLB->ISACencLB_obj.pitchfiltstr_obj); + WebRtcIsac_InitPitchAnalysis( + &instLB->ISACencLB_obj.pitchanalysisstr_obj); + + instLB->ISACencLB_obj.buffer_index = 0; + instLB->ISACencLB_obj.frame_nb = 0; + /* Default for I-mode. */ + instLB->ISACencLB_obj.bottleneck = 32000; + instLB->ISACencLB_obj.current_framesamples = 0; + instLB->ISACencLB_obj.s2nr = 0; + instLB->ISACencLB_obj.payloadLimitBytes30 = STREAM_SIZE_MAX_30; + instLB->ISACencLB_obj.payloadLimitBytes60 = STREAM_SIZE_MAX_60; + instLB->ISACencLB_obj.maxPayloadBytes = STREAM_SIZE_MAX_60; + instLB->ISACencLB_obj.maxRateInBytes = STREAM_SIZE_MAX_30; + instLB->ISACencLB_obj.enforceFrameSize = 0; + /* Invalid value prevents getRedPayload to + run before encoder is called. */ + instLB->ISACencLB_obj.lastBWIdx = -1; + return statusInit; +} + +static int16_t EncoderInitUb(ISACUBStruct* instUB, + int16_t bandwidth) { + int16_t statusInit = 0; + int k; + + /* Init stream vector to zero. */ + for (k = 0; k < STREAM_SIZE_MAX_60; k++) { + instUB->ISACencUB_obj.bitstr_obj.stream[k] = 0; + } + + WebRtcIsac_InitMasking(&instUB->ISACencUB_obj.maskfiltstr_obj); + WebRtcIsac_InitPreFilterbank(&instUB->ISACencUB_obj.prefiltbankstr_obj); + + if (bandwidth == isac16kHz) { + instUB->ISACencUB_obj.buffer_index = LB_TOTAL_DELAY_SAMPLES; + } else { + instUB->ISACencUB_obj.buffer_index = 0; + } + /* Default for I-mode. */ + instUB->ISACencUB_obj.bottleneck = 32000; + /* These store the limits for the wideband + super-wideband bit-stream. */ + instUB->ISACencUB_obj.maxPayloadSizeBytes = STREAM_SIZE_MAX_30 << 1; + /* This has to be updated after each lower-band encoding to guarantee + * a correct payload-limitation. */ + instUB->ISACencUB_obj.numBytesUsed = 0; + memset(instUB->ISACencUB_obj.data_buffer_float, 0, + (MAX_FRAMESAMPLES + LB_TOTAL_DELAY_SAMPLES) * sizeof(float)); + + memcpy(&(instUB->ISACencUB_obj.lastLPCVec), + WebRtcIsac_kMeanLarUb16, sizeof(double) * UB_LPC_ORDER); + + return statusInit; +} + + +int16_t WebRtcIsac_EncoderInit(ISACStruct* ISAC_main_inst, + int16_t codingMode) { + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + int16_t status; + + if ((codingMode != 0) && (codingMode != 1)) { + instISAC->errorCode = ISAC_DISALLOWED_CODING_MODE; + return -1; + } + /* Default bottleneck. */ + instISAC->bottleneck = MAX_ISAC_BW; + + if (instISAC->encoderSamplingRateKHz == kIsacWideband) { + instISAC->bandwidthKHz = isac8kHz; + instISAC->maxPayloadSizeBytes = STREAM_SIZE_MAX_60; + instISAC->maxRateBytesPer30Ms = STREAM_SIZE_MAX_30; + } else { + instISAC->bandwidthKHz = isac16kHz; + instISAC->maxPayloadSizeBytes = STREAM_SIZE_MAX; + instISAC->maxRateBytesPer30Ms = STREAM_SIZE_MAX; + } + + /* Channel-adaptive = 0; Instantaneous (Channel-independent) = 1. */ + instISAC->codingMode = codingMode; + + WebRtcIsac_InitBandwidthEstimator(&instISAC->bwestimator_obj, + instISAC->encoderSamplingRateKHz, + instISAC->decoderSamplingRateKHz); + + WebRtcIsac_InitRateModel(&instISAC->rate_data_obj); + /* Default for I-mode. */ + instISAC->MaxDelay = 10.0; + + status = EncoderInitLb(&instISAC->instLB, codingMode, + instISAC->encoderSamplingRateKHz); + if (status < 0) { + instISAC->errorCode = -status; + return -1; + } + + if (instISAC->encoderSamplingRateKHz == kIsacSuperWideband) { + /* Initialize encoder filter-bank. */ + memset(instISAC->analysisFBState1, 0, + FB_STATE_SIZE_WORD32 * sizeof(int32_t)); + memset(instISAC->analysisFBState2, 0, + FB_STATE_SIZE_WORD32 * sizeof(int32_t)); + + status = EncoderInitUb(&(instISAC->instUB), + instISAC->bandwidthKHz); + if (status < 0) { + instISAC->errorCode = -status; + return -1; + } + } + /* Initialization is successful, set the flag. */ + instISAC->initFlag |= BIT_MASK_ENC_INIT; + return 0; +} + + +/**************************************************************************** + * WebRtcIsac_Encode(...) + * + * This function encodes 10ms frame(s) and inserts it into a package. + * Input speech length has to be 160 samples (10ms). The encoder buffers those + * 10ms frames until it reaches the chosen Framesize (480 or 960 samples + * corresponding to 30 or 60 ms frames), and then proceeds to the encoding. + * + * Input: + * - ISAC_main_inst : ISAC instance. + * - speechIn : input speech vector. + * + * Output: + * - encoded : the encoded data vector + * + * Return value: + * : >0 - Length (in bytes) of coded data + * : 0 - The buffer didn't reach the chosen + * frameSize so it keeps buffering speech + * samples. + * : -1 - Error + */ +int WebRtcIsac_Encode(ISACStruct* ISAC_main_inst, + const int16_t* speechIn, + uint8_t* encoded) { + float inFrame[FRAMESAMPLES_10ms]; + int16_t speechInLB[FRAMESAMPLES_10ms]; + int16_t speechInUB[FRAMESAMPLES_10ms]; + int streamLenLB = 0; + int streamLenUB = 0; + int streamLen = 0; + size_t k = 0; + uint8_t garbageLen = 0; + int32_t bottleneck = 0; + int16_t bottleneckIdx = 0; + int16_t jitterInfo = 0; + + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + ISACLBStruct* instLB = &(instISAC->instLB); + ISACUBStruct* instUB = &(instISAC->instUB); + + /* Check if encoder initiated. */ + if ((instISAC->initFlag & BIT_MASK_ENC_INIT) != + BIT_MASK_ENC_INIT) { + instISAC->errorCode = ISAC_ENCODER_NOT_INITIATED; + return -1; + } + + if (instISAC->encoderSamplingRateKHz == kIsacSuperWideband) { + WebRtcSpl_AnalysisQMF(speechIn, SWBFRAMESAMPLES_10ms, speechInLB, + speechInUB, instISAC->analysisFBState1, + instISAC->analysisFBState2); + + /* Convert from fixed to floating point. */ + for (k = 0; k < FRAMESAMPLES_10ms; k++) { + inFrame[k] = (float)speechInLB[k]; + } + } else { + for (k = 0; k < FRAMESAMPLES_10ms; k++) { + inFrame[k] = (float) speechIn[k]; + } + } + + /* Add some noise to avoid denormal numbers. */ + inFrame[0] += (float)1.23455334e-3; + inFrame[1] -= (float)2.04324239e-3; + inFrame[2] += (float)1.90854954e-3; + inFrame[9] += (float)1.84854878e-3; + + /* This function will update the bottleneck if required. */ + UpdateBottleneck(instISAC); + + /* Get the bandwith information which has to be sent to the other side. */ + GetSendBandwidthInfo(instISAC, &bottleneckIdx, &jitterInfo); + + /* Encode lower-band. */ + streamLenLB = WebRtcIsac_EncodeLb(&instISAC->transform_tables, + inFrame, &instLB->ISACencLB_obj, + instISAC->codingMode, bottleneckIdx); + if (streamLenLB < 0) { + return -1; + } + + if (instISAC->encoderSamplingRateKHz == kIsacSuperWideband) { + instUB = &(instISAC->instUB); + + /* Convert to float. */ + for (k = 0; k < FRAMESAMPLES_10ms; k++) { + inFrame[k] = (float) speechInUB[k]; + } + + /* Add some noise to avoid denormal numbers. */ + inFrame[0] += (float)1.23455334e-3; + inFrame[1] -= (float)2.04324239e-3; + inFrame[2] += (float)1.90854954e-3; + inFrame[9] += (float)1.84854878e-3; + + /* Tell to upper-band the number of bytes used so far. + * This is for payload limitation. */ + instUB->ISACencUB_obj.numBytesUsed = + (int16_t)(streamLenLB + 1 + LEN_CHECK_SUM_WORD8); + /* Encode upper-band. */ + switch (instISAC->bandwidthKHz) { + case isac12kHz: { + streamLenUB = WebRtcIsac_EncodeUb12(&instISAC->transform_tables, + inFrame, &instUB->ISACencUB_obj, + jitterInfo); + break; + } + case isac16kHz: { + streamLenUB = WebRtcIsac_EncodeUb16(&instISAC->transform_tables, + inFrame, &instUB->ISACencUB_obj, + jitterInfo); + break; + } + case isac8kHz: { + streamLenUB = 0; + break; + } + } + + if ((streamLenUB < 0) && (streamLenUB != -ISAC_PAYLOAD_LARGER_THAN_LIMIT)) { + /* An error has happened but this is not the error due to a + * bit-stream larger than the limit. */ + return -1; + } + + if (streamLenLB == 0) { + return 0; + } + + /* One byte is allocated for the length. According to older decoders + so the length bit-stream plus one byte for size and + LEN_CHECK_SUM_WORD8 for the checksum should be less than or equal + to 255. */ + if ((streamLenUB > (255 - (LEN_CHECK_SUM_WORD8 + 1))) || + (streamLenUB == -ISAC_PAYLOAD_LARGER_THAN_LIMIT)) { + /* We have got a too long bit-stream we skip the upper-band + * bit-stream for this frame. */ + streamLenUB = 0; + } + + memcpy(encoded, instLB->ISACencLB_obj.bitstr_obj.stream, streamLenLB); + streamLen = streamLenLB; + if (streamLenUB > 0) { + encoded[streamLenLB] = (uint8_t)(streamLenUB + 1 + LEN_CHECK_SUM_WORD8); + memcpy(&encoded[streamLenLB + 1], + instUB->ISACencUB_obj.bitstr_obj.stream, + streamLenUB); + streamLen += encoded[streamLenLB]; + } else { + encoded[streamLenLB] = 0; + } + } else { + if (streamLenLB == 0) { + return 0; + } + memcpy(encoded, instLB->ISACencLB_obj.bitstr_obj.stream, streamLenLB); + streamLenUB = 0; + streamLen = streamLenLB; + } + + /* Add Garbage if required. */ + bottleneck = WebRtcIsac_GetUplinkBandwidth(&instISAC->bwestimator_obj); + if (instISAC->codingMode == 0) { + int minBytes; + int limit; + uint8_t* ptrGarbage; + + instISAC->MaxDelay = (double)WebRtcIsac_GetUplinkMaxDelay( + &instISAC->bwestimator_obj); + + /* Update rate model and get minimum number of bytes in this packet. */ + minBytes = WebRtcIsac_GetMinBytes( + &(instISAC->rate_data_obj), streamLen, + instISAC->instLB.ISACencLB_obj.current_framesamples, bottleneck, + instISAC->MaxDelay, instISAC->bandwidthKHz); + + /* Make sure MinBytes does not exceed packet size limit. */ + if (instISAC->bandwidthKHz == isac8kHz) { + if (instLB->ISACencLB_obj.current_framesamples == FRAMESAMPLES) { + limit = instLB->ISACencLB_obj.payloadLimitBytes30; + } else { + limit = instLB->ISACencLB_obj.payloadLimitBytes60; + } + } else { + limit = instUB->ISACencUB_obj.maxPayloadSizeBytes; + } + minBytes = (minBytes > limit) ? limit : minBytes; + + /* Make sure we don't allow more than 255 bytes of garbage data. + * We store the length of the garbage data in 8 bits in the bitstream, + * 255 is the max garbage length we can signal using 8 bits. */ + if ((instISAC->bandwidthKHz == isac8kHz) || + (streamLenUB == 0)) { + ptrGarbage = &encoded[streamLenLB]; + limit = streamLen + 255; + } else { + ptrGarbage = &encoded[streamLenLB + 1 + streamLenUB]; + limit = streamLen + (255 - encoded[streamLenLB]); + } + minBytes = (minBytes > limit) ? limit : minBytes; + + garbageLen = (minBytes > streamLen) ? (uint8_t)(minBytes - streamLen) : 0; + + /* Save data for creation of multiple bit-streams. */ + /* If bit-stream too short then add garbage at the end. */ + if (garbageLen > 0) { + /* Overwrite the garbage area to avoid leaking possibly sensitive data + over the network. This also makes the output deterministic. */ + memset(ptrGarbage, 0, garbageLen); + + /* For a correct length of the upper-band bit-stream together + * with the garbage. Garbage is embeded in upper-band bit-stream. + * That is the only way to preserve backward compatibility. */ + if ((instISAC->bandwidthKHz == isac8kHz) || + (streamLenUB == 0)) { + encoded[streamLenLB] = garbageLen; + } else { + encoded[streamLenLB] += garbageLen; + /* Write the length of the garbage at the end of the upper-band + * bit-stream, if exists. This helps for sanity check. */ + encoded[streamLenLB + 1 + streamLenUB] = garbageLen; + + } + streamLen += garbageLen; + } + } else { + /* update rate model */ + WebRtcIsac_UpdateRateModel( + &instISAC->rate_data_obj, streamLen, + instISAC->instLB.ISACencLB_obj.current_framesamples, bottleneck); + garbageLen = 0; + } + + /* Generate CRC if required. */ + if ((instISAC->bandwidthKHz != isac8kHz) && (streamLenUB > 0)) { + uint32_t crc; + + WebRtcIsac_GetCrc((int16_t*)(&(encoded[streamLenLB + 1])), + streamLenUB + garbageLen, &crc); +#ifndef WEBRTC_ARCH_BIG_ENDIAN + for (k = 0; k < LEN_CHECK_SUM_WORD8; k++) { + encoded[streamLen - LEN_CHECK_SUM_WORD8 + k] = + (uint8_t)(crc >> (24 - k * 8)); + } +#else + memcpy(&encoded[streamLenLB + streamLenUB + 1], &crc, LEN_CHECK_SUM_WORD8); +#endif + } + return streamLen; +} + + +/****************************************************************************** + * WebRtcIsac_GetNewBitStream(...) + * + * This function returns encoded data, with the recieved bwe-index in the + * stream. If the rate is set to a value less than bottleneck of codec + * the new bistream will be re-encoded with the given target rate. + * It should always return a complete packet, i.e. only called once + * even for 60 msec frames. + * + * NOTE 1! This function does not write in the ISACStruct, it is not allowed. + * NOTE 2! Rates larger than the bottleneck of the codec will be limited + * to the current bottleneck. + * + * Input: + * - ISAC_main_inst : ISAC instance. + * - bweIndex : Index of bandwidth estimate to put in new + * bitstream + * - rate : target rate of the transcoder is bits/sec. + * Valid values are the accepted rate in iSAC, + * i.e. 10000 to 56000. + * + * Output: + * - encoded : The encoded data vector + * + * Return value : >0 - Length (in bytes) of coded data + * -1 - Error or called in SWB mode + * NOTE! No error code is written to + * the struct since it is only allowed to read + * the struct. + */ +int16_t WebRtcIsac_GetNewBitStream(ISACStruct* ISAC_main_inst, + int16_t bweIndex, + int16_t jitterInfo, + int32_t rate, + uint8_t* encoded, + int16_t isRCU) { + Bitstr iSACBitStreamInst; /* Local struct for bitstream handling */ + int16_t streamLenLB; + int16_t streamLenUB; + int16_t totalStreamLen; + double gain2; + double gain1; + float scale; + enum ISACBandwidth bandwidthKHz; + double rateLB; + double rateUB; + int32_t currentBN; + uint32_t crc; +#ifndef WEBRTC_ARCH_BIG_ENDIAN + int16_t k; +#endif + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + + if ((instISAC->initFlag & BIT_MASK_ENC_INIT) != + BIT_MASK_ENC_INIT) { + return -1; + } + + /* Get the bottleneck of this iSAC and limit the + * given rate to the current bottleneck. */ + WebRtcIsac_GetUplinkBw(ISAC_main_inst, ¤tBN); + if (rate > currentBN) { + rate = currentBN; + } + + if (WebRtcIsac_RateAllocation(rate, &rateLB, &rateUB, &bandwidthKHz) < 0) { + return -1; + } + + /* Cannot transcode from 16 kHz to 12 kHz. */ + if ((bandwidthKHz == isac12kHz) && + (instISAC->bandwidthKHz == isac16kHz)) { + return -1; + } + + /* A gain [dB] for the given rate. */ + gain1 = WebRtcIsac_GetSnr( + rateLB, instISAC->instLB.ISACencLB_obj.current_framesamples); + /* The gain [dB] of this iSAC. */ + gain2 = WebRtcIsac_GetSnr( + instISAC->instLB.ISACencLB_obj.bottleneck, + instISAC->instLB.ISACencLB_obj.current_framesamples); + + /* Scale is the ratio of two gains in normal domain. */ + scale = (float)pow(10, (gain1 - gain2) / 20.0); + /* Change the scale if this is a RCU bit-stream. */ + scale = (isRCU) ? (scale * RCU_TRANSCODING_SCALE) : scale; + + streamLenLB = WebRtcIsac_EncodeStoredDataLb( + &instISAC->instLB.ISACencLB_obj.SaveEnc_obj, + &iSACBitStreamInst, bweIndex, scale); + + if (streamLenLB < 0) { + return -1; + } + + /* Convert from bytes to int16_t. */ + memcpy(encoded, iSACBitStreamInst.stream, streamLenLB); + + if (bandwidthKHz == isac8kHz) { + return streamLenLB; + } + + totalStreamLen = streamLenLB; + /* super-wideband is always at 30ms. + * These gains are in dB. + * Gain for the given rate. */ + gain1 = WebRtcIsac_GetSnr(rateUB, FRAMESAMPLES); + /* Gain of this iSAC */ + gain2 = WebRtcIsac_GetSnr(instISAC->instUB.ISACencUB_obj.bottleneck, + FRAMESAMPLES); + + /* Scale is the ratio of two gains in normal domain. */ + scale = (float)pow(10, (gain1 - gain2) / 20.0); + + /* Change the scale if this is a RCU bit-stream. */ + scale = (isRCU)? (scale * RCU_TRANSCODING_SCALE_UB) : scale; + + streamLenUB = WebRtcIsac_EncodeStoredDataUb( + &(instISAC->instUB.ISACencUB_obj.SaveEnc_obj), + &iSACBitStreamInst, jitterInfo, scale, + instISAC->bandwidthKHz); + + if (streamLenUB < 0) { + return -1; + } + + if (streamLenUB + 1 + LEN_CHECK_SUM_WORD8 > 255) { + return streamLenLB; + } + + totalStreamLen = streamLenLB + streamLenUB + 1 + LEN_CHECK_SUM_WORD8; + encoded[streamLenLB] = streamLenUB + 1 + LEN_CHECK_SUM_WORD8; + + memcpy(&encoded[streamLenLB + 1], iSACBitStreamInst.stream, + streamLenUB); + + WebRtcIsac_GetCrc((int16_t*)(&(encoded[streamLenLB + 1])), + streamLenUB, &crc); +#ifndef WEBRTC_ARCH_BIG_ENDIAN + for (k = 0; k < LEN_CHECK_SUM_WORD8; k++) { + encoded[totalStreamLen - LEN_CHECK_SUM_WORD8 + k] = + (uint8_t)((crc >> (24 - k * 8)) & 0xFF); + } +#else + memcpy(&encoded[streamLenLB + streamLenUB + 1], &crc, + LEN_CHECK_SUM_WORD8); +#endif + return totalStreamLen; +} + + +/**************************************************************************** + * DecoderInitLb(...) - internal function for initialization of + * Lower Band + * DecoderInitUb(...) - internal function for initialization of + * Upper Band + * WebRtcIsac_DecoderInit(...) - API function + * + * This function initializes a ISAC instance prior to the decoder calls. + * + * Input: + * - ISAC_main_inst : ISAC instance. + */ +static void DecoderInitLb(ISACLBStruct* instISAC) { + int i; + /* Initialize stream vector to zero. */ + for (i = 0; i < STREAM_SIZE_MAX_60; i++) { + instISAC->ISACdecLB_obj.bitstr_obj.stream[i] = 0; + } + + WebRtcIsac_InitMasking(&instISAC->ISACdecLB_obj.maskfiltstr_obj); + WebRtcIsac_InitPostFilterbank( + &instISAC->ISACdecLB_obj.postfiltbankstr_obj); + WebRtcIsac_InitPitchFilter(&instISAC->ISACdecLB_obj.pitchfiltstr_obj); +} + +static void DecoderInitUb(ISACUBStruct* instISAC) { + int i; + /* Init stream vector to zero */ + for (i = 0; i < STREAM_SIZE_MAX_60; i++) { + instISAC->ISACdecUB_obj.bitstr_obj.stream[i] = 0; + } + + WebRtcIsac_InitMasking(&instISAC->ISACdecUB_obj.maskfiltstr_obj); + WebRtcIsac_InitPostFilterbank( + &instISAC->ISACdecUB_obj.postfiltbankstr_obj); +} + +void WebRtcIsac_DecoderInit(ISACStruct* ISAC_main_inst) { + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + + DecoderInitLb(&instISAC->instLB); + if (instISAC->decoderSamplingRateKHz == kIsacSuperWideband) { + memset(instISAC->synthesisFBState1, 0, + FB_STATE_SIZE_WORD32 * sizeof(int32_t)); + memset(instISAC->synthesisFBState2, 0, + FB_STATE_SIZE_WORD32 * sizeof(int32_t)); + DecoderInitUb(&(instISAC->instUB)); + } + if ((instISAC->initFlag & BIT_MASK_ENC_INIT) != BIT_MASK_ENC_INIT) { + WebRtcIsac_InitBandwidthEstimator(&instISAC->bwestimator_obj, + instISAC->encoderSamplingRateKHz, + instISAC->decoderSamplingRateKHz); + } + instISAC->initFlag |= BIT_MASK_DEC_INIT; + instISAC->resetFlag_8kHz = 0; +} + + +/**************************************************************************** + * WebRtcIsac_UpdateBwEstimate(...) + * + * This function updates the estimate of the bandwidth. + * + * NOTE: + * The estimates of bandwidth is not valid if the sample rate of the far-end + * encoder is set to 48 kHz and send timestamps are increamented according to + * 48 kHz sampling rate. + * + * Input: + * - ISAC_main_inst : ISAC instance. + * - encoded : encoded ISAC frame(s). + * - packet_size : size of the packet. + * - rtp_seq_number : the RTP number of the packet. + * - arr_ts : the arrival time of the packet (from NetEq) + * in samples. + * + * Return value : 0 - Ok + * -1 - Error + */ +int16_t WebRtcIsac_UpdateBwEstimate(ISACStruct* ISAC_main_inst, + const uint8_t* encoded, + size_t packet_size, + uint16_t rtp_seq_number, + uint32_t send_ts, + uint32_t arr_ts) { + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + Bitstr streamdata; +#ifndef WEBRTC_ARCH_BIG_ENDIAN + int k; +#endif + int16_t err; + + /* Check if decoder initiated. */ + if ((instISAC->initFlag & BIT_MASK_DEC_INIT) != BIT_MASK_DEC_INIT) { + instISAC->errorCode = ISAC_DECODER_NOT_INITIATED; + return -1; + } + + /* Check that the size of the packet is valid, and if not return without + * updating the bandwidth estimate. A valid size is at least 10 bytes. */ + if (packet_size < 10) { + /* Return error code if the packet length is null. */ + instISAC->errorCode = ISAC_EMPTY_PACKET; + return -1; + } + + WebRtcIsac_ResetBitstream(&(streamdata)); + +#ifndef WEBRTC_ARCH_BIG_ENDIAN + for (k = 0; k < 10; k++) { + uint16_t ek = ((const uint16_t*)encoded)[k >> 1]; + streamdata.stream[k] = (uint8_t)((ek >> ((k & 1) << 3)) & 0xff); + } +#else + memcpy(streamdata.stream, encoded, 10); +#endif + + err = WebRtcIsac_EstimateBandwidth(&instISAC->bwestimator_obj, &streamdata, + packet_size, rtp_seq_number, send_ts, + arr_ts, instISAC->encoderSamplingRateKHz, + instISAC->decoderSamplingRateKHz); + if (err < 0) { + /* Return error code if something went wrong. */ + instISAC->errorCode = -err; + return -1; + } + return 0; +} + +static int Decode(ISACStruct* ISAC_main_inst, + const uint8_t* encoded, + size_t lenEncodedBytes, + int16_t* decoded, + int16_t* speechType, + int16_t isRCUPayload) { + /* Number of samples (480 or 960), output from decoder + that were actually used in the encoder/decoder + (determined on the fly). */ + int16_t numSamplesLB; + int16_t numSamplesUB; + int16_t speechIdx; + float outFrame[MAX_FRAMESAMPLES]; + int16_t outFrameLB[MAX_FRAMESAMPLES]; + int16_t outFrameUB[MAX_FRAMESAMPLES]; + int numDecodedBytesLBint; + size_t numDecodedBytesLB; + int numDecodedBytesUB; + size_t lenEncodedLBBytes; + int16_t validChecksum = 1; + int16_t k; + uint16_t numLayer; + size_t totSizeBytes; + int16_t err; + + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + ISACUBDecStruct* decInstUB = &(instISAC->instUB.ISACdecUB_obj); + ISACLBDecStruct* decInstLB = &(instISAC->instLB.ISACdecLB_obj); + + /* Check if decoder initiated. */ + if ((instISAC->initFlag & BIT_MASK_DEC_INIT) != + BIT_MASK_DEC_INIT) { + instISAC->errorCode = ISAC_DECODER_NOT_INITIATED; + return -1; + } + + if (lenEncodedBytes == 0) { + /* return error code if the packet length is null. */ + instISAC->errorCode = ISAC_EMPTY_PACKET; + return -1; + } + + /* The size of the encoded lower-band is bounded by + * STREAM_SIZE_MAX. If a payload with the size larger than STREAM_SIZE_MAX + * is received, it is not considered erroneous. */ + lenEncodedLBBytes = (lenEncodedBytes > STREAM_SIZE_MAX) ? + STREAM_SIZE_MAX : lenEncodedBytes; + + /* Copy to lower-band bit-stream structure. */ + memcpy(instISAC->instLB.ISACdecLB_obj.bitstr_obj.stream, encoded, + lenEncodedLBBytes); + + /* We need to initialize numSamplesLB to something; otherwise, in the test + for whether we should return -1 below, the compiler might generate code + that fools Memcheck (Valgrind) into thinking that the control flow depends + on the uninitialized value in numSamplesLB (since WebRtcIsac_DecodeLb will + not fill it in if it fails and returns -1). */ + numSamplesLB = 0; + + /* Regardless of that the current codec is setup to work in + * wideband or super-wideband, the decoding of the lower-band + * has to be performed. */ + numDecodedBytesLBint = WebRtcIsac_DecodeLb(&instISAC->transform_tables, + outFrame, decInstLB, + &numSamplesLB, isRCUPayload); + numDecodedBytesLB = (size_t)numDecodedBytesLBint; + if ((numDecodedBytesLBint < 0) || + (numDecodedBytesLB > lenEncodedLBBytes) || + (numSamplesLB > MAX_FRAMESAMPLES)) { + instISAC->errorCode = ISAC_LENGTH_MISMATCH; + return -1; + } + + /* Error Check, we accept multi-layer bit-stream This will limit number + * of iterations of the while loop. Even without this the number + * of iterations is limited. */ + numLayer = 1; + totSizeBytes = numDecodedBytesLB; + while (totSizeBytes != lenEncodedBytes) { + if ((totSizeBytes > lenEncodedBytes) || + (encoded[totSizeBytes] == 0) || + (numLayer > MAX_NUM_LAYERS)) { + instISAC->errorCode = ISAC_LENGTH_MISMATCH; + return -1; + } + totSizeBytes += encoded[totSizeBytes]; + numLayer++; + } + + if (instISAC->decoderSamplingRateKHz == kIsacWideband) { + for (k = 0; k < numSamplesLB; k++) { + if (outFrame[k] > 32767) { + decoded[k] = 32767; + } else if (outFrame[k] < -32768) { + decoded[k] = -32768; + } else { + decoded[k] = (int16_t)WebRtcIsac_lrint(outFrame[k]); + } + } + numSamplesUB = 0; + } else { + uint32_t crc; + /* We don't accept larger than 30ms (480 samples at lower-band) + * frame-size. */ + for (k = 0; k < numSamplesLB; k++) { + if (outFrame[k] > 32767) { + outFrameLB[k] = 32767; + } else if (outFrame[k] < -32768) { + outFrameLB[k] = -32768; + } else { + outFrameLB[k] = (int16_t)WebRtcIsac_lrint(outFrame[k]); + } + } + + /* Check for possible error, and if upper-band stream exists. */ + if (numDecodedBytesLB == lenEncodedBytes) { + /* Decoding was successful. No super-wideband bit-stream exists. */ + numSamplesUB = numSamplesLB; + memset(outFrameUB, 0, sizeof(int16_t) * numSamplesUB); + + /* Prepare for the potential increase of signal bandwidth. */ + instISAC->resetFlag_8kHz = 2; + } else { + /* This includes the checksum and the bytes that stores the length. */ + int16_t lenNextStream = encoded[numDecodedBytesLB]; + + /* Is this garbage or valid super-wideband bit-stream? + * Check if checksum is valid. */ + if (lenNextStream <= (LEN_CHECK_SUM_WORD8 + 1)) { + /* Such a small second layer cannot be super-wideband layer. + * It must be a short garbage. */ + validChecksum = 0; + } else { + /* Run CRC to see if the checksum match. */ + WebRtcIsac_GetCrc((int16_t*)(&encoded[numDecodedBytesLB + 1]), + lenNextStream - LEN_CHECK_SUM_WORD8 - 1, &crc); + + validChecksum = 1; + for (k = 0; k < LEN_CHECK_SUM_WORD8; k++) { + validChecksum &= (((crc >> (24 - k * 8)) & 0xFF) == + encoded[numDecodedBytesLB + lenNextStream - + LEN_CHECK_SUM_WORD8 + k]); + } + } + + if (!validChecksum) { + /* This is a garbage, we have received a wideband + * bit-stream with garbage. */ + numSamplesUB = numSamplesLB; + memset(outFrameUB, 0, sizeof(int16_t) * numSamplesUB); + } else { + /* A valid super-wideband biststream exists. */ + enum ISACBandwidth bandwidthKHz; + int32_t maxDelayBit; + + /* If we have super-wideband bit-stream, we cannot + * have 60 ms frame-size. */ + if (numSamplesLB > FRAMESAMPLES) { + instISAC->errorCode = ISAC_LENGTH_MISMATCH; + return -1; + } + + /* The rest of the bit-stream contains the upper-band + * bit-stream curently this is the only thing there, + * however, we might add more layers. */ + + /* Have to exclude one byte where the length is stored + * and last 'LEN_CHECK_SUM_WORD8' bytes where the + * checksum is stored. */ + lenNextStream -= (LEN_CHECK_SUM_WORD8 + 1); + + memcpy(decInstUB->bitstr_obj.stream, + &encoded[numDecodedBytesLB + 1], lenNextStream); + + /* Reset bit-stream object, this is the first decoding. */ + WebRtcIsac_ResetBitstream(&(decInstUB->bitstr_obj)); + + /* Decode jitter information. */ + err = WebRtcIsac_DecodeJitterInfo(&decInstUB->bitstr_obj, &maxDelayBit); + if (err < 0) { + instISAC->errorCode = -err; + return -1; + } + + /* Update jitter info which is in the upper-band bit-stream + * only if the encoder is in super-wideband. Otherwise, + * the jitter info is already embedded in bandwidth index + * and has been updated. */ + if (instISAC->encoderSamplingRateKHz == kIsacSuperWideband) { + err = WebRtcIsac_UpdateUplinkJitter( + &(instISAC->bwestimator_obj), maxDelayBit); + if (err < 0) { + instISAC->errorCode = -err; + return -1; + } + } + + /* Decode bandwidth information. */ + err = WebRtcIsac_DecodeBandwidth(&decInstUB->bitstr_obj, + &bandwidthKHz); + if (err < 0) { + instISAC->errorCode = -err; + return -1; + } + + switch (bandwidthKHz) { + case isac12kHz: { + numDecodedBytesUB = WebRtcIsac_DecodeUb12( + &instISAC->transform_tables, outFrame, decInstUB, isRCUPayload); + + /* Hang-over for transient alleviation - + * wait two frames to add the upper band going up from 8 kHz. */ + if (instISAC->resetFlag_8kHz > 0) { + if (instISAC->resetFlag_8kHz == 2) { + /* Silence first and a half frame. */ + memset(outFrame, 0, MAX_FRAMESAMPLES * + sizeof(float)); + } else { + const float rampStep = 2.0f / MAX_FRAMESAMPLES; + float rampVal = 0; + memset(outFrame, 0, (MAX_FRAMESAMPLES >> 1) * + sizeof(float)); + + /* Ramp up second half of second frame. */ + for (k = MAX_FRAMESAMPLES / 2; k < MAX_FRAMESAMPLES; k++) { + outFrame[k] *= rampVal; + rampVal += rampStep; + } + } + instISAC->resetFlag_8kHz -= 1; + } + + break; + } + case isac16kHz: { + numDecodedBytesUB = WebRtcIsac_DecodeUb16( + &instISAC->transform_tables, outFrame, decInstUB, isRCUPayload); + break; + } + default: + return -1; + } + + /* It might be less due to garbage. */ + if ((numDecodedBytesUB != lenNextStream) && + (numDecodedBytesUB != (lenNextStream - + encoded[numDecodedBytesLB + 1 + numDecodedBytesUB]))) { + instISAC->errorCode = ISAC_LENGTH_MISMATCH; + return -1; + } + + /* If there is no error Upper-band always decodes + * 30 ms (480 samples). */ + numSamplesUB = FRAMESAMPLES; + + /* Convert to W16. */ + for (k = 0; k < numSamplesUB; k++) { + if (outFrame[k] > 32767) { + outFrameUB[k] = 32767; + } else if (outFrame[k] < -32768) { + outFrameUB[k] = -32768; + } else { + outFrameUB[k] = (int16_t)WebRtcIsac_lrint( + outFrame[k]); + } + } + } + } + + speechIdx = 0; + while (speechIdx < numSamplesLB) { + WebRtcSpl_SynthesisQMF(&outFrameLB[speechIdx], &outFrameUB[speechIdx], + FRAMESAMPLES_10ms, &decoded[(speechIdx << 1)], + instISAC->synthesisFBState1, + instISAC->synthesisFBState2); + + speechIdx += FRAMESAMPLES_10ms; + } + } + *speechType = 0; + return (numSamplesLB + numSamplesUB); +} + + + + + + + +/**************************************************************************** + * WebRtcIsac_Decode(...) + * + * This function decodes a ISAC frame. Output speech length + * will be a multiple of 480 samples: 480 or 960 samples, + * depending on the frameSize (30 or 60 ms). + * + * Input: + * - ISAC_main_inst : ISAC instance. + * - encoded : encoded ISAC frame(s) + * - len : bytes in encoded vector + * + * Output: + * - decoded : The decoded vector + * + * Return value : >0 - number of samples in decoded vector + * -1 - Error + */ + +int WebRtcIsac_Decode(ISACStruct* ISAC_main_inst, + const uint8_t* encoded, + size_t lenEncodedBytes, + int16_t* decoded, + int16_t* speechType) { + int16_t isRCUPayload = 0; + return Decode(ISAC_main_inst, encoded, lenEncodedBytes, decoded, + speechType, isRCUPayload); +} + +/**************************************************************************** + * WebRtcIsac_DecodeRcu(...) + * + * This function decodes a redundant (RCU) iSAC frame. Function is called in + * NetEq with a stored RCU payload in case of packet loss. Output speech length + * will be a multiple of 480 samples: 480 or 960 samples, + * depending on the framesize (30 or 60 ms). + * + * Input: + * - ISAC_main_inst : ISAC instance. + * - encoded : encoded ISAC RCU frame(s) + * - len : bytes in encoded vector + * + * Output: + * - decoded : The decoded vector + * + * Return value : >0 - number of samples in decoded vector + * -1 - Error + */ + + + +int WebRtcIsac_DecodeRcu(ISACStruct* ISAC_main_inst, + const uint8_t* encoded, + size_t lenEncodedBytes, + int16_t* decoded, + int16_t* speechType) { + int16_t isRCUPayload = 1; + return Decode(ISAC_main_inst, encoded, lenEncodedBytes, decoded, + speechType, isRCUPayload); +} + + +/**************************************************************************** + * WebRtcIsac_DecodePlc(...) + * + * This function conducts PLC for ISAC frame(s). Output speech length + * will be a multiple of 480 samples: 480 or 960 samples, + * depending on the frameSize (30 or 60 ms). + * + * Input: + * - ISAC_main_inst : ISAC instance. + * - noOfLostFrames : Number of PLC frames to produce + * + * Output: + * - decoded : The decoded vector + * + * Return value : Number of samples in decoded PLC vector + */ +size_t WebRtcIsac_DecodePlc(ISACStruct* ISAC_main_inst, + int16_t* decoded, + size_t noOfLostFrames) { + size_t numSamples = 0; + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + + /* Limit number of frames to two = 60 millisecond. + * Otherwise we exceed data vectors. */ + if (noOfLostFrames > 2) { + noOfLostFrames = 2; + } + + /* Get the number of samples per frame */ + switch (instISAC->decoderSamplingRateKHz) { + case kIsacWideband: { + numSamples = 480 * noOfLostFrames; + break; + } + case kIsacSuperWideband: { + numSamples = 960 * noOfLostFrames; + break; + } + } + + /* Set output samples to zero. */ + memset(decoded, 0, numSamples * sizeof(int16_t)); + return numSamples; +} + + +/**************************************************************************** + * ControlLb(...) - Internal function for controlling Lower Band + * ControlUb(...) - Internal function for controlling Upper Band + * WebRtcIsac_Control(...) - API function + * + * This function sets the limit on the short-term average bit rate and the + * frame length. Should be used only in Instantaneous mode. + * + * Input: + * - ISAC_main_inst : ISAC instance. + * - rate : limit on the short-term average bit rate, + * in bits/second (between 10000 and 32000) + * - frameSize : number of milliseconds per frame (30 or 60) + * + * Return value : 0 - ok + * -1 - Error + */ +static int16_t ControlLb(ISACLBStruct* instISAC, double rate, + int16_t frameSize) { + if ((rate >= 10000) && (rate <= 32000)) { + instISAC->ISACencLB_obj.bottleneck = rate; + } else { + return -ISAC_DISALLOWED_BOTTLENECK; + } + + if ((frameSize == 30) || (frameSize == 60)) { + instISAC->ISACencLB_obj.new_framelength = (FS / 1000) * frameSize; + } else { + return -ISAC_DISALLOWED_FRAME_LENGTH; + } + + return 0; +} + +static int16_t ControlUb(ISACUBStruct* instISAC, double rate) { + if ((rate >= 10000) && (rate <= 32000)) { + instISAC->ISACencUB_obj.bottleneck = rate; + } else { + return -ISAC_DISALLOWED_BOTTLENECK; + } + return 0; +} + +int16_t WebRtcIsac_Control(ISACStruct* ISAC_main_inst, + int32_t bottleneckBPS, + int frameSize) { + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + int16_t status; + double rateLB; + double rateUB; + enum ISACBandwidth bandwidthKHz; + + if (instISAC->codingMode == 0) { + /* In adaptive mode. */ + instISAC->errorCode = ISAC_MODE_MISMATCH; + return -1; + } + + /* Check if encoder initiated */ + if ((instISAC->initFlag & BIT_MASK_ENC_INIT) != + BIT_MASK_ENC_INIT) { + instISAC->errorCode = ISAC_ENCODER_NOT_INITIATED; + return -1; + } + + if (instISAC->encoderSamplingRateKHz == kIsacWideband) { + /* If the sampling rate is 16kHz then bandwith should be 8kHz, + * regardless of bottleneck. */ + bandwidthKHz = isac8kHz; + rateLB = (bottleneckBPS > 32000) ? 32000 : bottleneckBPS; + rateUB = 0; + } else { + if (WebRtcIsac_RateAllocation(bottleneckBPS, &rateLB, &rateUB, + &bandwidthKHz) < 0) { + return -1; + } + } + + if ((instISAC->encoderSamplingRateKHz == kIsacSuperWideband) && + (frameSize != 30) && + (bandwidthKHz != isac8kHz)) { + /* Cannot have 60 ms in super-wideband. */ + instISAC->errorCode = ISAC_DISALLOWED_FRAME_LENGTH; + return -1; + } + + status = ControlLb(&instISAC->instLB, rateLB, (int16_t)frameSize); + if (status < 0) { + instISAC->errorCode = -status; + return -1; + } + if (bandwidthKHz != isac8kHz) { + status = ControlUb(&(instISAC->instUB), rateUB); + if (status < 0) { + instISAC->errorCode = -status; + return -1; + } + } + + + /* Check if bandwidth is changing from wideband to super-wideband + * then we have to synch data buffer of lower & upper-band. Also + * clean up the upper-band data buffer. */ + + if ((instISAC->bandwidthKHz == isac8kHz) && (bandwidthKHz != isac8kHz)) { + memset(instISAC->instUB.ISACencUB_obj.data_buffer_float, 0, + sizeof(float) * (MAX_FRAMESAMPLES + LB_TOTAL_DELAY_SAMPLES)); + + if (bandwidthKHz == isac12kHz) { + instISAC->instUB.ISACencUB_obj.buffer_index = + instISAC->instLB.ISACencLB_obj.buffer_index; + } else { + instISAC->instUB.ISACencUB_obj.buffer_index = + LB_TOTAL_DELAY_SAMPLES + instISAC->instLB.ISACencLB_obj.buffer_index; + + memcpy(&(instISAC->instUB.ISACencUB_obj.lastLPCVec), + WebRtcIsac_kMeanLarUb16, sizeof(double) * UB_LPC_ORDER); + } + } + + /* Update the payload limit if the bandwidth is changing. */ + if (instISAC->bandwidthKHz != bandwidthKHz) { + instISAC->bandwidthKHz = bandwidthKHz; + UpdatePayloadSizeLimit(instISAC); + } + instISAC->bottleneck = bottleneckBPS; + return 0; +} + +void WebRtcIsac_SetInitialBweBottleneck(ISACStruct* ISAC_main_inst, + int bottleneck_bits_per_second) { + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + assert(bottleneck_bits_per_second >= 10000 && + bottleneck_bits_per_second <= 32000); + instISAC->bwestimator_obj.send_bw_avg = (float)bottleneck_bits_per_second; +} + +/**************************************************************************** + * WebRtcIsac_ControlBwe(...) + * + * This function sets the initial values of bottleneck and frame-size if + * iSAC is used in channel-adaptive mode. Through this API, users can + * enforce a frame-size for all values of bottleneck. Then iSAC will not + * automatically change the frame-size. + * + * + * Input: + * - ISAC_main_inst : ISAC instance. + * - rateBPS : initial value of bottleneck in bits/second + * 10000 <= rateBPS <= 32000 is accepted + * For default bottleneck set rateBPS = 0 + * - frameSizeMs : number of milliseconds per frame (30 or 60) + * - enforceFrameSize : 1 to enforce the given frame-size through out + * the adaptation process, 0 to let iSAC change + * the frame-size if required. + * + * Return value : 0 - ok + * -1 - Error + */ +int16_t WebRtcIsac_ControlBwe(ISACStruct* ISAC_main_inst, + int32_t bottleneckBPS, + int frameSizeMs, + int16_t enforceFrameSize) { + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + enum ISACBandwidth bandwidth; + + /* Check if encoder initiated */ + if ((instISAC->initFlag & BIT_MASK_ENC_INIT) != + BIT_MASK_ENC_INIT) { + instISAC->errorCode = ISAC_ENCODER_NOT_INITIATED; + return -1; + } + + /* Check that we are in channel-adaptive mode, otherwise, return (-1) */ + if (instISAC->codingMode != 0) { + instISAC->errorCode = ISAC_MODE_MISMATCH; + return -1; + } + if ((frameSizeMs != 30) && + (instISAC->encoderSamplingRateKHz == kIsacSuperWideband)) { + return -1; + } + + /* Set structure variable if enforceFrameSize is set. ISAC will then + * keep the chosen frame size. */ + if (enforceFrameSize != 0) { + instISAC->instLB.ISACencLB_obj.enforceFrameSize = 1; + } else { + instISAC->instLB.ISACencLB_obj.enforceFrameSize = 0; + } + + /* Set the initial rate. If the input value is zero then the default intial + * rate is used. Otehrwise, values between 10 to 32 kbps are accepted. */ + if (bottleneckBPS != 0) { + double rateLB; + double rateUB; + if (WebRtcIsac_RateAllocation(bottleneckBPS, &rateLB, &rateUB, + &bandwidth) < 0) { + return -1; + } + instISAC->bwestimator_obj.send_bw_avg = (float)bottleneckBPS; + instISAC->bandwidthKHz = bandwidth; + } + + /* Set the initial frame-size. If 'enforceFrameSize' is set, the frame-size + * will not change */ + if (frameSizeMs != 0) { + if ((frameSizeMs == 30) || (frameSizeMs == 60)) { + instISAC->instLB.ISACencLB_obj.new_framelength = + (int16_t)((FS / 1000) * frameSizeMs); + } else { + instISAC->errorCode = ISAC_DISALLOWED_FRAME_LENGTH; + return -1; + } + } + return 0; +} + + +/**************************************************************************** + * WebRtcIsac_GetDownLinkBwIndex(...) + * + * This function returns index representing the Bandwidth estimate from + * the other side to this side. + * + * Input: + * - ISAC_main_inst : iSAC structure + * + * Output: + * - bweIndex : Bandwidth estimate to transmit to other side. + * + */ +int16_t WebRtcIsac_GetDownLinkBwIndex(ISACStruct* ISAC_main_inst, + int16_t* bweIndex, + int16_t* jitterInfo) { + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + + /* Check if encoder initialized. */ + if ((instISAC->initFlag & BIT_MASK_DEC_INIT) != + BIT_MASK_DEC_INIT) { + instISAC->errorCode = ISAC_ENCODER_NOT_INITIATED; + return -1; + } + + /* Call function to get Bandwidth Estimate. */ + WebRtcIsac_GetDownlinkBwJitIndexImpl(&(instISAC->bwestimator_obj), bweIndex, + jitterInfo, + instISAC->decoderSamplingRateKHz); + return 0; +} + + +/**************************************************************************** + * WebRtcIsac_UpdateUplinkBw(...) + * + * This function takes an index representing the Bandwidth estimate from + * this side to other side and updates BWE. + * + * Input: + * - ISAC_main_inst : iSAC structure + * - rateIndex : Bandwidth estimate from other side. + * + * Return value : 0 - ok + * -1 - index out of range + */ +int16_t WebRtcIsac_UpdateUplinkBw(ISACStruct* ISAC_main_inst, + int16_t bweIndex) { + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + int16_t returnVal; + + /* Check if encoder initiated. */ + if ((instISAC->initFlag & BIT_MASK_ENC_INIT) != + BIT_MASK_ENC_INIT) { + instISAC->errorCode = ISAC_ENCODER_NOT_INITIATED; + return -1; + } + + /* Call function to get Bandwidth Estimate. */ + returnVal = WebRtcIsac_UpdateUplinkBwImpl( + &(instISAC->bwestimator_obj), bweIndex, + instISAC->encoderSamplingRateKHz); + + if (returnVal < 0) { + instISAC->errorCode = -returnVal; + return -1; + } else { + return 0; + } +} + + +/**************************************************************************** + * WebRtcIsac_ReadBwIndex(...) + * + * This function returns the index of the Bandwidth estimate from the + * bit-stream. + * + * Input: + * - encoded : Encoded bit-stream + * + * Output: + * - frameLength : Length of frame in packet (in samples) + * - bweIndex : Bandwidth estimate in bit-stream + * + */ +int16_t WebRtcIsac_ReadBwIndex(const uint8_t* encoded, + int16_t* bweIndex) { + Bitstr streamdata; +#ifndef WEBRTC_ARCH_BIG_ENDIAN + int k; +#endif + int16_t err; + + WebRtcIsac_ResetBitstream(&(streamdata)); + +#ifndef WEBRTC_ARCH_BIG_ENDIAN + for (k = 0; k < 10; k++) { + int16_t ek2 = ((const int16_t*)encoded)[k >> 1]; + streamdata.stream[k] = (uint8_t)((ek2 >> ((k & 1) << 3)) & 0xff); + } +#else + memcpy(streamdata.stream, encoded, 10); +#endif + + /* Decode frame length. */ + err = WebRtcIsac_DecodeFrameLen(&streamdata, bweIndex); + if (err < 0) { + return err; + } + + /* Decode BW estimation. */ + err = WebRtcIsac_DecodeSendBW(&streamdata, bweIndex); + if (err < 0) { + return err; + } + + return 0; +} + + +/**************************************************************************** + * WebRtcIsac_ReadFrameLen(...) + * + * This function returns the number of samples the decoder will generate if + * the given payload is decoded. + * + * Input: + * - encoded : Encoded bitstream + * + * Output: + * - frameLength : Length of frame in packet (in samples) + * + */ +int16_t WebRtcIsac_ReadFrameLen(ISACStruct* ISAC_main_inst, + const uint8_t* encoded, + int16_t* frameLength) { + Bitstr streamdata; +#ifndef WEBRTC_ARCH_BIG_ENDIAN + int k; +#endif + int16_t err; + ISACMainStruct* instISAC; + + WebRtcIsac_ResetBitstream(&(streamdata)); + +#ifndef WEBRTC_ARCH_BIG_ENDIAN + for (k = 0; k < 10; k++) { + int16_t ek2 = ((const int16_t*)encoded)[k >> 1]; + streamdata.stream[k] = (uint8_t)((ek2 >> ((k & 1) << 3)) & 0xff); + } +#else + memcpy(streamdata.stream, encoded, 10); +#endif + + /* Decode frame length. */ + err = WebRtcIsac_DecodeFrameLen(&streamdata, frameLength); + if (err < 0) { + return -1; + } + instISAC = (ISACMainStruct*)ISAC_main_inst; + + if (instISAC->decoderSamplingRateKHz == kIsacSuperWideband) { + /* The decoded frame length indicates the number of samples in + * lower-band in this case, multiply by 2 to get the total number + * of samples. */ + *frameLength <<= 1; + } + return 0; +} + + +/******************************************************************************* + * WebRtcIsac_GetNewFrameLen(...) + * + * This function returns the frame length (in samples) of the next packet. + * In the case of channel-adaptive mode, iSAC decides on its frame length based + * on the estimated bottleneck, this AOI allows a user to prepare for the next + * packet (at the encoder). + * + * The primary usage is in CE to make the iSAC works in channel-adaptive mode + * + * Input: + * - ISAC_main_inst : iSAC struct + * + * Return Value : frame lenght in samples + * + */ +int16_t WebRtcIsac_GetNewFrameLen(ISACStruct* ISAC_main_inst) { + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + + /* Return new frame length. */ + if (instISAC->in_sample_rate_hz == 16000) + return (instISAC->instLB.ISACencLB_obj.new_framelength); + else /* 32000 Hz */ + return ((instISAC->instLB.ISACencLB_obj.new_framelength) * 2); +} + + +/**************************************************************************** + * WebRtcIsac_GetErrorCode(...) + * + * This function can be used to check the error code of an iSAC instance. + * When a function returns -1 an error code will be set for that instance. + * The function below extracts the code of the last error that occurred in + * the specified instance. + * + * Input: + * - ISAC_main_inst : ISAC instance + * + * Return value : Error code + */ +int16_t WebRtcIsac_GetErrorCode(ISACStruct* ISAC_main_inst) { + return ((ISACMainStruct*)ISAC_main_inst)->errorCode; +} + + +/**************************************************************************** + * WebRtcIsac_GetUplinkBw(...) + * + * This function outputs the target bottleneck of the codec. In + * channel-adaptive mode, the target bottleneck is specified through an in-band + * signalling retrieved by bandwidth estimator. + * In channel-independent, also called instantaneous mode, the target + * bottleneck is provided to the encoder by calling xxx_control(...) (if + * xxx_control is never called, the default values are used.). + * Note that the output is the iSAC internal operating bottleneck which might + * differ slightly from the one provided through xxx_control(). + * + * Input: + * - ISAC_main_inst : iSAC instance + * + * Output: + * - *bottleneck : bottleneck in bits/sec + * + * Return value : -1 if error happens + * 0 bit-rates computed correctly. + */ +int16_t WebRtcIsac_GetUplinkBw(ISACStruct* ISAC_main_inst, + int32_t* bottleneck) { + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + + if (instISAC->codingMode == 0) { + /* We are in adaptive mode then get the bottleneck from BWE. */ + *bottleneck = (int32_t)instISAC->bwestimator_obj.send_bw_avg; + } else { + *bottleneck = instISAC->bottleneck; + } + + if ((*bottleneck > 32000) && (*bottleneck < 38000)) { + *bottleneck = 32000; + } else if ((*bottleneck > 45000) && (*bottleneck < 50000)) { + *bottleneck = 45000; + } else if (*bottleneck > 56000) { + *bottleneck = 56000; + } + return 0; +} + + +/****************************************************************************** + * WebRtcIsac_SetMaxPayloadSize(...) + * + * This function sets a limit for the maximum payload size of iSAC. The same + * value is used both for 30 and 60 ms packets. If the encoder sampling rate + * is 16 kHz the maximum payload size is between 120 and 400 bytes. If the + * encoder sampling rate is 32 kHz the maximum payload size is between 120 + * and 600 bytes. + * + * --------------- + * IMPORTANT NOTES + * --------------- + * The size of a packet is limited to the minimum of 'max-payload-size' and + * 'max-rate.' For instance, let's assume the max-payload-size is set to + * 170 bytes, and max-rate is set to 40 kbps. Note that a limit of 40 kbps + * translates to 150 bytes for 30ms frame-size & 300 bytes for 60ms + * frame-size. Then a packet with a frame-size of 30 ms is limited to 150, + * i.e. min(170, 150), and a packet with 60 ms frame-size is limited to + * 170 bytes, i.e. min(170, 300). + * + * Input: + * - ISAC_main_inst : iSAC instance + * - maxPayloadBytes : maximum size of the payload in bytes + * valid values are between 100 and 400 bytes + * if encoder sampling rate is 16 kHz. For + * 32 kHz encoder sampling rate valid values + * are between 100 and 600 bytes. + * + * Return value : 0 if successful + * -1 if error happens + */ +int16_t WebRtcIsac_SetMaxPayloadSize(ISACStruct* ISAC_main_inst, + int16_t maxPayloadBytes) { + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + int16_t status = 0; + + /* Check if encoder initiated */ + if ((instISAC->initFlag & BIT_MASK_ENC_INIT) != + BIT_MASK_ENC_INIT) { + instISAC->errorCode = ISAC_ENCODER_NOT_INITIATED; + return -1; + } + + if (instISAC->encoderSamplingRateKHz == kIsacSuperWideband) { + /* Sanity check. */ + if (maxPayloadBytes < 120) { + /* 'maxRate' is out of valid range + * set to the acceptable value and return -1. */ + maxPayloadBytes = 120; + status = -1; + } + + /* sanity check */ + if (maxPayloadBytes > STREAM_SIZE_MAX) { + /* maxRate is out of valid range, + * set to the acceptable value and return -1. */ + maxPayloadBytes = STREAM_SIZE_MAX; + status = -1; + } + } else { + if (maxPayloadBytes < 120) { + /* Max payload-size is out of valid range + * set to the acceptable value and return -1. */ + maxPayloadBytes = 120; + status = -1; + } + if (maxPayloadBytes > STREAM_SIZE_MAX_60) { + /* Max payload-size is out of valid range + * set to the acceptable value and return -1. */ + maxPayloadBytes = STREAM_SIZE_MAX_60; + status = -1; + } + } + instISAC->maxPayloadSizeBytes = maxPayloadBytes; + UpdatePayloadSizeLimit(instISAC); + return status; +} + + +/****************************************************************************** + * WebRtcIsac_SetMaxRate(...) + * + * This function sets the maximum rate which the codec may not exceed for + * any signal packet. The maximum rate is defined and payload-size per + * frame-size in bits per second. + * + * The codec has a maximum rate of 53400 bits per second (200 bytes per 30 + * ms) if the encoder sampling rate is 16kHz, and 160 kbps (600 bytes/30 ms) + * if the encoder sampling rate is 32 kHz. + * + * It is possible to set a maximum rate between 32000 and 53400 bits/sec + * in wideband mode, and 32000 to 160000 bits/sec in super-wideband mode. + * + * --------------- + * IMPORTANT NOTES + * --------------- + * The size of a packet is limited to the minimum of 'max-payload-size' and + * 'max-rate.' For instance, let's assume the max-payload-size is set to + * 170 bytes, and max-rate is set to 40 kbps. Note that a limit of 40 kbps + * translates to 150 bytes for 30ms frame-size & 300 bytes for 60ms + * frame-size. Then a packet with a frame-size of 30 ms is limited to 150, + * i.e. min(170, 150), and a packet with 60 ms frame-size is limited to + * 170 bytes, min(170, 300). + * + * Input: + * - ISAC_main_inst : iSAC instance + * - maxRate : maximum rate in bits per second, + * valid values are 32000 to 53400 bits/sec in + * wideband mode, and 32000 to 160000 bits/sec in + * super-wideband mode. + * + * Return value : 0 if successful + * -1 if error happens + */ +int16_t WebRtcIsac_SetMaxRate(ISACStruct* ISAC_main_inst, + int32_t maxRate) { + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + int16_t maxRateInBytesPer30Ms; + int16_t status = 0; + + /* check if encoder initiated */ + if ((instISAC->initFlag & BIT_MASK_ENC_INIT) != BIT_MASK_ENC_INIT) { + instISAC->errorCode = ISAC_ENCODER_NOT_INITIATED; + return -1; + } + /* Calculate maximum number of bytes per 30 msec packets for the + given maximum rate. Multiply with 30/1000 to get number of + bits per 30 ms, divide by 8 to get number of bytes per 30 ms: + maxRateInBytes = floor((maxRate * 30/1000) / 8); */ + maxRateInBytesPer30Ms = (int16_t)(maxRate * 3 / 800); + + if (instISAC->encoderSamplingRateKHz == kIsacWideband) { + if (maxRate < 32000) { + /* 'maxRate' is out of valid range. + * Set to the acceptable value and return -1. */ + maxRateInBytesPer30Ms = 120; + status = -1; + } + + if (maxRate > 53400) { + /* 'maxRate' is out of valid range. + * Set to the acceptable value and return -1. */ + maxRateInBytesPer30Ms = 200; + status = -1; + } + } else { + if (maxRateInBytesPer30Ms < 120) { + /* 'maxRate' is out of valid range + * Set to the acceptable value and return -1. */ + maxRateInBytesPer30Ms = 120; + status = -1; + } + + if (maxRateInBytesPer30Ms > STREAM_SIZE_MAX) { + /* 'maxRate' is out of valid range. + * Set to the acceptable value and return -1. */ + maxRateInBytesPer30Ms = STREAM_SIZE_MAX; + status = -1; + } + } + instISAC->maxRateBytesPer30Ms = maxRateInBytesPer30Ms; + UpdatePayloadSizeLimit(instISAC); + return status; +} + + +/**************************************************************************** + * WebRtcIsac_GetRedPayload(...) + * + * This function populates "encoded" with the redundant payload of the recently + * encodedframe. This function has to be called once that WebRtcIsac_Encode(...) + * returns a positive value. Regardless of the frame-size this function will + * be called only once after encoding is completed. The bit-stream is + * targeted for 16000 bit/sec. + * + * Input: + * - ISAC_main_inst : iSAC struct + * + * Output: + * - encoded : the encoded data vector + * + * + * Return value : >0 - Length (in bytes) of coded data + * : -1 - Error + */ +int16_t WebRtcIsac_GetRedPayload(ISACStruct* ISAC_main_inst, + uint8_t* encoded) { + Bitstr iSACBitStreamInst; + int16_t streamLenLB; + int16_t streamLenUB; + int16_t streamLen; + int16_t totalLenUB; + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; +#ifndef WEBRTC_ARCH_BIG_ENDIAN + int k; +#endif + + if ((instISAC->initFlag & BIT_MASK_ENC_INIT) != + BIT_MASK_ENC_INIT) { + instISAC->errorCode = ISAC_ENCODER_NOT_INITIATED; + } + + WebRtcIsac_ResetBitstream(&(iSACBitStreamInst)); + + streamLenLB = WebRtcIsac_EncodeStoredDataLb( + &instISAC->instLB.ISACencLB_obj.SaveEnc_obj, + &iSACBitStreamInst, + instISAC->instLB.ISACencLB_obj.lastBWIdx, + RCU_TRANSCODING_SCALE); + if (streamLenLB < 0) { + return -1; + } + + /* convert from bytes to int16_t. */ + memcpy(encoded, iSACBitStreamInst.stream, streamLenLB); + streamLen = streamLenLB; + if (instISAC->bandwidthKHz == isac8kHz) { + return streamLenLB; + } + + streamLenUB = WebRtcIsac_GetRedPayloadUb( + &instISAC->instUB.ISACencUB_obj.SaveEnc_obj, + &iSACBitStreamInst, instISAC->bandwidthKHz); + if (streamLenUB < 0) { + /* An error has happened but this is not the error due to a + * bit-stream larger than the limit. */ + return -1; + } + + /* We have one byte to write the total length of the upper-band. + * The length includes the bit-stream length, check-sum and the + * single byte where the length is written to. This is according to + * iSAC wideband and how the "garbage" is dealt. */ + totalLenUB = streamLenUB + 1 + LEN_CHECK_SUM_WORD8; + if (totalLenUB > 255) { + streamLenUB = 0; + } + + /* Generate CRC if required. */ + if ((instISAC->bandwidthKHz != isac8kHz) && + (streamLenUB > 0)) { + uint32_t crc; + streamLen += totalLenUB; + encoded[streamLenLB] = (uint8_t)totalLenUB; + memcpy(&encoded[streamLenLB + 1], iSACBitStreamInst.stream, + streamLenUB); + + WebRtcIsac_GetCrc((int16_t*)(&(encoded[streamLenLB + 1])), + streamLenUB, &crc); +#ifndef WEBRTC_ARCH_BIG_ENDIAN + for (k = 0; k < LEN_CHECK_SUM_WORD8; k++) { + encoded[streamLen - LEN_CHECK_SUM_WORD8 + k] = + (uint8_t)((crc >> (24 - k * 8)) & 0xFF); + } +#else + memcpy(&encoded[streamLenLB + streamLenUB + 1], &crc, + LEN_CHECK_SUM_WORD8); +#endif + } + return streamLen; +} + + +/**************************************************************************** + * WebRtcIsac_version(...) + * + * This function returns the version number. + * + * Output: + * - version : Pointer to character string + * + */ +void WebRtcIsac_version(char* version) { + strcpy(version, "4.3.0"); +} + + +/****************************************************************************** + * WebRtcIsac_SetEncSampRate() + * This function sets the sampling rate of the encoder. Initialization of the + * encoder WILL NOT overwrite the sampling rate of the encoder. The default + * value is 16 kHz which is set when the instance is created. The encoding-mode + * and the bottleneck remain unchanged by this call, however, the maximum rate + * and maximum payload-size will be reset to their default values. + * + * Input: + * - ISAC_main_inst : iSAC instance + * - sample_rate_hz : sampling rate in Hertz, valid values are 16000 + * and 32000. + * + * Return value : 0 if successful + * -1 if failed. + */ +int16_t WebRtcIsac_SetEncSampRate(ISACStruct* ISAC_main_inst, + uint16_t sample_rate_hz) { + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + enum IsacSamplingRate encoder_operational_rate; + + if ((sample_rate_hz != 16000) && (sample_rate_hz != 32000)) { + /* Sampling Frequency is not supported. */ + instISAC->errorCode = ISAC_UNSUPPORTED_SAMPLING_FREQUENCY; + return -1; + } + if (sample_rate_hz == 16000) { + encoder_operational_rate = kIsacWideband; + } else { + encoder_operational_rate = kIsacSuperWideband; + } + + if ((instISAC->initFlag & BIT_MASK_ENC_INIT) != + BIT_MASK_ENC_INIT) { + if (encoder_operational_rate == kIsacWideband) { + instISAC->bandwidthKHz = isac8kHz; + } else { + instISAC->bandwidthKHz = isac16kHz; + } + } else { + ISACUBStruct* instUB = &(instISAC->instUB); + ISACLBStruct* instLB = &(instISAC->instLB); + int32_t bottleneck = instISAC->bottleneck; + int16_t codingMode = instISAC->codingMode; + int16_t frameSizeMs = instLB->ISACencLB_obj.new_framelength / + (FS / 1000); + + if ((encoder_operational_rate == kIsacWideband) && + (instISAC->encoderSamplingRateKHz == kIsacSuperWideband)) { + /* Changing from super-wideband to wideband. + * we don't need to re-initialize the encoder of the lower-band. */ + instISAC->bandwidthKHz = isac8kHz; + if (codingMode == 1) { + ControlLb(instLB, + (bottleneck > 32000) ? 32000 : bottleneck, FRAMESIZE); + } + instISAC->maxPayloadSizeBytes = STREAM_SIZE_MAX_60; + instISAC->maxRateBytesPer30Ms = STREAM_SIZE_MAX_30; + } else if ((encoder_operational_rate == kIsacSuperWideband) && + (instISAC->encoderSamplingRateKHz == kIsacWideband)) { + double bottleneckLB = 0; + double bottleneckUB = 0; + if (codingMode == 1) { + WebRtcIsac_RateAllocation(bottleneck, &bottleneckLB, &bottleneckUB, + &(instISAC->bandwidthKHz)); + } + + instISAC->bandwidthKHz = isac16kHz; + instISAC->maxPayloadSizeBytes = STREAM_SIZE_MAX; + instISAC->maxRateBytesPer30Ms = STREAM_SIZE_MAX; + + EncoderInitLb(instLB, codingMode, encoder_operational_rate); + EncoderInitUb(instUB, instISAC->bandwidthKHz); + + memset(instISAC->analysisFBState1, 0, + FB_STATE_SIZE_WORD32 * sizeof(int32_t)); + memset(instISAC->analysisFBState2, 0, + FB_STATE_SIZE_WORD32 * sizeof(int32_t)); + + if (codingMode == 1) { + instISAC->bottleneck = bottleneck; + ControlLb(instLB, bottleneckLB, + (instISAC->bandwidthKHz == isac8kHz) ? frameSizeMs:FRAMESIZE); + if (instISAC->bandwidthKHz > isac8kHz) { + ControlUb(instUB, bottleneckUB); + } + } else { + instLB->ISACencLB_obj.enforceFrameSize = 0; + instLB->ISACencLB_obj.new_framelength = FRAMESAMPLES; + } + } + } + instISAC->encoderSamplingRateKHz = encoder_operational_rate; + instISAC->in_sample_rate_hz = sample_rate_hz; + return 0; +} + + +/****************************************************************************** + * WebRtcIsac_SetDecSampRate() + * This function sets the sampling rate of the decoder. Initialization of the + * decoder WILL NOT overwrite the sampling rate of the encoder. The default + * value is 16 kHz which is set when the instance is created. + * + * Input: + * - ISAC_main_inst : iSAC instance + * - sample_rate_hz : sampling rate in Hertz, valid values are 16000 + * and 32000. + * + * Return value : 0 if successful + * -1 if failed. + */ +int16_t WebRtcIsac_SetDecSampRate(ISACStruct* ISAC_main_inst, + uint16_t sample_rate_hz) { + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + enum IsacSamplingRate decoder_operational_rate; + + if (sample_rate_hz == 16000) { + decoder_operational_rate = kIsacWideband; + } else if (sample_rate_hz == 32000) { + decoder_operational_rate = kIsacSuperWideband; + } else { + /* Sampling Frequency is not supported. */ + instISAC->errorCode = ISAC_UNSUPPORTED_SAMPLING_FREQUENCY; + return -1; + } + + if ((instISAC->decoderSamplingRateKHz == kIsacWideband) && + (decoder_operational_rate == kIsacSuperWideband)) { + /* Switching from wideband to super-wideband at the decoder + * we need to reset the filter-bank and initialize upper-band decoder. */ + memset(instISAC->synthesisFBState1, 0, + FB_STATE_SIZE_WORD32 * sizeof(int32_t)); + memset(instISAC->synthesisFBState2, 0, + FB_STATE_SIZE_WORD32 * sizeof(int32_t)); + + DecoderInitUb(&instISAC->instUB); + } + instISAC->decoderSamplingRateKHz = decoder_operational_rate; + return 0; +} + + +/****************************************************************************** + * WebRtcIsac_EncSampRate() + * + * Input: + * - ISAC_main_inst : iSAC instance + * + * Return value : sampling rate in Hertz. The input to encoder + * is expected to be sampled in this rate. + * + */ +uint16_t WebRtcIsac_EncSampRate(ISACStruct* ISAC_main_inst) { + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + return instISAC->in_sample_rate_hz; +} + + +/****************************************************************************** + * WebRtcIsac_DecSampRate() + * Return the sampling rate of the decoded audio. + * + * Input: + * - ISAC_main_inst : iSAC instance + * + * Return value : sampling rate in Hertz. Decoder output is + * sampled at this rate. + * + */ +uint16_t WebRtcIsac_DecSampRate(ISACStruct* ISAC_main_inst) { + ISACMainStruct* instISAC = (ISACMainStruct*)ISAC_main_inst; + return instISAC->decoderSamplingRateKHz == kIsacWideband ? 16000 : 32000; +} + +void WebRtcIsac_GetBandwidthInfo(ISACStruct* inst, + IsacBandwidthInfo* bwinfo) { + ISACMainStruct* instISAC = (ISACMainStruct*)inst; + assert(instISAC->initFlag & BIT_MASK_DEC_INIT); + WebRtcIsacBw_GetBandwidthInfo(&instISAC->bwestimator_obj, + instISAC->decoderSamplingRateKHz, bwinfo); +} + +void WebRtcIsac_SetBandwidthInfo(ISACStruct* inst, + const IsacBandwidthInfo* bwinfo) { + ISACMainStruct* instISAC = (ISACMainStruct*)inst; + assert(instISAC->initFlag & BIT_MASK_ENC_INIT); + WebRtcIsacBw_SetBandwidthInfo(&instISAC->bwestimator_obj, bwinfo); +} + +void WebRtcIsac_SetEncSampRateInDecoder(ISACStruct* inst, + int sample_rate_hz) { + ISACMainStruct* instISAC = (ISACMainStruct*)inst; + assert(instISAC->initFlag & BIT_MASK_DEC_INIT); + assert(!(instISAC->initFlag & BIT_MASK_ENC_INIT)); + assert(sample_rate_hz == 16000 || sample_rate_hz == 32000); + instISAC->encoderSamplingRateKHz = sample_rate_hz / 1000; +} diff --git a/webrtc/modules/audio_coding/codecs/isac/main/source/isac_float_type.h b/webrtc/modules/audio_coding/codecs/isac/main/source/isac_float_type.h new file mode 100644 index 0000000..e150d39 --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/main/source/isac_float_type.h @@ -0,0 +1,117 @@ +/* + * Copyright (c) 2015 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#ifndef WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_SOURCE_ISAC_FLOAT_TYPE_H_ +#define WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_SOURCE_ISAC_FLOAT_TYPE_H_ + +#include "webrtc/modules/audio_coding/codecs/isac/main/include/isac.h" + +namespace webrtc { + +struct IsacFloat { + using instance_type = ISACStruct; + static const bool has_swb = true; + static inline int16_t Control(instance_type* inst, + int32_t rate, + int framesize) { + return WebRtcIsac_Control(inst, rate, framesize); + } + static inline int16_t ControlBwe(instance_type* inst, + int32_t rate_bps, + int frame_size_ms, + int16_t enforce_frame_size) { + return WebRtcIsac_ControlBwe(inst, rate_bps, frame_size_ms, + enforce_frame_size); + } + static inline int16_t Create(instance_type** inst) { + return WebRtcIsac_Create(inst); + } + static inline int DecodeInternal(instance_type* inst, + const uint8_t* encoded, + size_t len, + int16_t* decoded, + int16_t* speech_type) { + return WebRtcIsac_Decode(inst, encoded, len, decoded, speech_type); + } + static inline size_t DecodePlc(instance_type* inst, + int16_t* decoded, + size_t num_lost_frames) { + return WebRtcIsac_DecodePlc(inst, decoded, num_lost_frames); + } + + static inline void DecoderInit(instance_type* inst) { + WebRtcIsac_DecoderInit(inst); + } + static inline int Encode(instance_type* inst, + const int16_t* speech_in, + uint8_t* encoded) { + return WebRtcIsac_Encode(inst, speech_in, encoded); + } + static inline int16_t EncoderInit(instance_type* inst, int16_t coding_mode) { + return WebRtcIsac_EncoderInit(inst, coding_mode); + } + static inline uint16_t EncSampRate(instance_type* inst) { + return WebRtcIsac_EncSampRate(inst); + } + + static inline int16_t Free(instance_type* inst) { + return WebRtcIsac_Free(inst); + } + static inline void GetBandwidthInfo(instance_type* inst, + IsacBandwidthInfo* bwinfo) { + WebRtcIsac_GetBandwidthInfo(inst, bwinfo); + } + static inline int16_t GetErrorCode(instance_type* inst) { + return WebRtcIsac_GetErrorCode(inst); + } + + static inline int16_t GetNewFrameLen(instance_type* inst) { + return WebRtcIsac_GetNewFrameLen(inst); + } + static inline void SetBandwidthInfo(instance_type* inst, + const IsacBandwidthInfo* bwinfo) { + WebRtcIsac_SetBandwidthInfo(inst, bwinfo); + } + static inline int16_t SetDecSampRate(instance_type* inst, + uint16_t sample_rate_hz) { + return WebRtcIsac_SetDecSampRate(inst, sample_rate_hz); + } + static inline int16_t SetEncSampRate(instance_type* inst, + uint16_t sample_rate_hz) { + return WebRtcIsac_SetEncSampRate(inst, sample_rate_hz); + } + static inline void SetEncSampRateInDecoder(instance_type* inst, + uint16_t sample_rate_hz) { + WebRtcIsac_SetEncSampRateInDecoder(inst, sample_rate_hz); + } + static inline void SetInitialBweBottleneck(instance_type* inst, + int bottleneck_bits_per_second) { + WebRtcIsac_SetInitialBweBottleneck(inst, bottleneck_bits_per_second); + } + static inline int16_t UpdateBwEstimate(instance_type* inst, + const uint8_t* encoded, + size_t packet_size, + uint16_t rtp_seq_number, + uint32_t send_ts, + uint32_t arr_ts) { + return WebRtcIsac_UpdateBwEstimate(inst, encoded, packet_size, + rtp_seq_number, send_ts, arr_ts); + } + static inline int16_t SetMaxPayloadSize(instance_type* inst, + int16_t max_payload_size_bytes) { + return WebRtcIsac_SetMaxPayloadSize(inst, max_payload_size_bytes); + } + static inline int16_t SetMaxRate(instance_type* inst, int32_t max_bit_rate) { + return WebRtcIsac_SetMaxRate(inst, max_bit_rate); + } +}; + +} // namespace webrtc +#endif // WEBRTC_MODULES_AUDIO_CODING_CODECS_ISAC_MAIN_SOURCE_ISAC_FLOAT_TYPE_H_ diff --git a/webrtc/modules/audio_coding/codecs/isac/main/source/lattice.c b/webrtc/modules/audio_coding/codecs/isac/main/source/lattice.c new file mode 100644 index 0000000..eabe708 --- /dev/null +++ b/webrtc/modules/audio_coding/codecs/isac/main/source/lattice.c @@ -0,0 +1,218 @@ +/* + * Copyright (c) 2011 The WebRTC project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +/* + * lattice.c + * + * contains the normalized lattice filter routines (MA and AR) for iSAC codec + * + */ +#include "settings.h" +#include "codec.h" + +#include +#include +#include +#ifdef WEBRTC_ANDROID +#include +#endif + +/* filter the signal using normalized lattice filter */ +/* MA filter */ +void WebRtcIsac_NormLatticeFilterMa(int orderCoef, + float *stateF, + float *stateG, + float *lat_in, + double *filtcoeflo, + double *lat_out) +{ + int n,k,i,u,temp1; + int ord_1 = orderCoef+1; + float sth[MAX_AR_MODEL_ORDER]; + float cth[MAX_AR_MODEL_ORDER]; + float inv_cth[MAX_AR_MODEL_ORDER]; + double a[MAX_AR_MODEL_ORDER+1]; + float f[MAX_AR_MODEL_ORDER+1][HALF_SUBFRAMELEN], g[MAX_AR_MODEL_ORDER+1][HALF_SUBFRAMELEN]; + float gain1; + + for (u=0;u=0;i--) //get the state of f&g for the first input, for all orders + { + ARf[i][0] = cth[i]*ARf[i+1][0] - sth[i]*stateG[i]; + ARg[i+1][0] = sth[i]*ARf[i+1][0] + cth[i]* stateG[i]; + } + ARg[0][0] = ARf[0][0]; + + for(n=0;n<(HALF_SUBFRAMELEN-1);n++) + { + for(k=orderCoef-1;k>=0;k--) + { + ARf[k][n+1] = cth[k]*ARf[k+1][n+1] - sth[k]*ARg[k][n]; + ARg[k+1][n+1] = sth[k]*ARf[k+1][n+1] + cth[k]* ARg[k][n]; + } + ARg[0][n+1] = ARf[0][n+1]; + } + + memcpy(lat_out+u * HALF_SUBFRAMELEN, &(ARf[0][0]), sizeof(float) * HALF_SUBFRAMELEN); + + /* cannot use memcpy in the following */ + for (i=0;i0; m--) + { + tmp_inv = 1.0f / cth2; + for (k=1; k<=m; k++) + { + tmp[k] = ((float)a[k] - sth[m] * (float)a[m-k+1]) * tmp_inv; + } + + for (k=1; k + +void WebRtcIsac_InitTransform(TransformTables* tables) { + int k; + double fact, phase; + + fact = PI / (FRAMESAMPLES_HALF); + phase = 0.0; + for (k = 0; k < FRAMESAMPLES_HALF; k++) { + tables->costab1[k] = cos(phase); + tables->sintab1[k] = sin(phase); + phase += fact; + } + + fact = PI * ((double) (FRAMESAMPLES_HALF - 1)) / ((double) FRAMESAMPLES_HALF); + phase = 0.5 * fact; + for (k = 0; k < FRAMESAMPLES_QUARTER; k++) { + tables->costab2[k] = cos(phase); + tables->sintab2[k] = sin(phase); + phase += fact; + } +} + +void WebRtcIsac_Time2Spec(const TransformTables* tables, + double* inre1, + double* inre2, + int16_t* outreQ7, + int16_t* outimQ7, + FFTstr* fftstr_obj) { + int k; + int dims[1]; + double tmp1r, tmp1i, xr, xi, yr, yi, fact; + double tmpre[FRAMESAMPLES_HALF], tmpim[FRAMESAMPLES_HALF]; + + + dims[0] = FRAMESAMPLES_HALF; + + + /* Multiply with complex exponentials and combine into one complex vector */ + fact = 0.5 / sqrt(FRAMESAMPLES_HALF); + for (k = 0; k < FRAMESAMPLES_HALF; k++) { + tmp1r = tables->costab1[k]; + tmp1i = tables->sintab1[k]; + tmpre[k] = (inre1[k] * tmp1r + inre2[k] * tmp1i) * fact; + tmpim[k] = (inre2[k] * tmp1r - inre1[k] * tmp1i) * fact; + } + + + /* Get DFT */ + WebRtcIsac_Fftns(1, dims, tmpre, tmpim, -1, 1.0, fftstr_obj); + + /* Use symmetry to separate into two complex vectors and center frames in time around zero */ + for (k = 0; k < FRAMESAMPLES_QUARTER; k++) { + xr = tmpre[k] + tmpre[FRAMESAMPLES_HALF - 1 - k]; + yi = -tmpre[k] + tmpre[FRAMESAMPLES_HALF - 1 - k]; + xi = tmpim[k] - tmpim[FRAMESAMPLES_HALF - 1 - k]; + yr = tmpim[k] + tmpim[FRAMESAMPLES_HALF - 1 - k]; + + tmp1r = tables->costab2[k]; + tmp1i = tables->sintab2[k]; + outreQ7[k] = (int16_t)WebRtcIsac_lrint((xr * tmp1r - xi * tmp1i) * 128.0); + outimQ7[k] = (int16_t)WebRtcIsac_lrint((xr * tmp1i + xi * tmp1r) * 128.0); + outreQ7[FRAMESAMPLES_HALF - 1 - k] = (int16_t)WebRtcIsac_lrint((-yr * tmp1i - yi * tmp1r) * 128.0); + outimQ7[FRAMESAMPLES_HALF - 1 - k] = (int16_t)WebRtcIsac_lrint((-yr * tmp1r + yi * tmp1i) * 128.0); + } +} + +void WebRtcIsac_Spec2time(const TransformTables* tables, + double* inre, + double* inim, + double* outre1, + double* outre2, + FFTstr* fftstr_obj) { + int k; + double tmp1r, tmp1i, xr, xi, yr, yi, fact; + + int dims; + + dims = FRAMESAMPLES_HALF; + + for (k = 0; k < FRAMESAMPLES_QUARTER; k++) { + /* Move zero in time to beginning of frames */ + tmp1r = tables->costab2[k]; + tmp1i = tables->sintab2[k]; + xr = inre[k] * tmp1r + inim[k] * tmp1i; + xi = inim[k] * tmp1r - inre[k] * tmp1i; + yr = -inim[FRAMESAMPLES_HALF - 1 - k] * tmp1r - inre[FRAMESAMPLES_HALF - 1 - k] * tmp1i; + yi = -inre[FRAMESAMPLES_HALF - 1 - k] * tmp1r + inim[FRAMESAMPLES_HALF - 1 - k] * tmp1i; + + /* Combine into one vector, z = x + j * y */ + outre1[k] = xr - yi; + outre1[FRAMESAMPLES_HALF - 1 - k] = xr + yi; + outre2[k] = xi + yr; + outre2[FRAMESAMPLES_HALF - 1 - k] = -xi + yr; + } + + + /* Get IDFT */ + WebRtcIsac_Fftns(1, &dims, outre1, outre2, 1, FRAMESAMPLES_HALF, fftstr_obj); + + + /* Demodulate and separate */ + fact = sqrt(FRAMESAMPLES_HALF); + for (k = 0; k < FRAMESAMPLES_HALF; k++) { + tmp1r = tables->costab1[k]; + tmp1i = tables->sintab1[k]; + xr = (outre1[k] * tmp1r - outre2[k] * tmp1i) * fact; + outre2[k] = (outre2[k] * tmp1r + outre1[k] * tmp1i) * fact; + outre1[k] = xr; + } +} diff --git a/webrtc/modules/audio_coding/meson.build b/webrtc/modules/audio_coding/meson.build index 5975bfa..fb46df8 100644 --- a/webrtc/modules/audio_coding/meson.build +++ b/webrtc/modules/audio_coding/meson.build @@ -1,4 +1,4 @@ -audio_coding_sources = [ +webrtc_audio_coding_sources = [ 'codecs/isac/main/source/arith_routines.c', 'codecs/isac/main/source/arith_routines_hist.c', 'codecs/isac/main/source/arith_routines_logist.c', @@ -18,16 +18,45 @@ audio_coding_sources = [ 'codecs/isac/main/source/pitch_gain_tables.c', 'codecs/isac/main/source/pitch_lag_tables.c', 'codecs/isac/main/source/spectrum_ar_model_tables.c', + 'codecs/audio_decoder.cc', + 'codecs/audio_encoder.cc', + 'codecs/isac/main/source/audio_decoder_isac.cc', + 'codecs/isac/main/source/audio_encoder_isac.cc', + 'codecs/isac/main/source/bandwidth_estimator.c', + 'codecs/isac/main/source/crc.c', + 'codecs/isac/main/source/decode.c', + 'codecs/isac/main/source/decode_bwe.c', + 'codecs/isac/main/source/encode.c', + 'codecs/isac/main/source/fft.c', + 'codecs/isac/main/source/isac.c', + 'codecs/isac/main/source/lattice.c', + 'codecs/isac/main/source/transform.c', ] -libaudio_coding = static_library('audio_coding', - audio_coding_sources, - dependencies: common_deps, +libwebrtc_audio_coding = library('webrtc_audio_coding', + webrtc_audio_coding_sources, + dependencies: [base_dep, common_audio_dep] + common_deps, include_directories: webrtc_inc, c_args: common_cflags, - cpp_args: common_cxxflags + cpp_args: common_cxxflags, + soversion: soversion, + install: true ) -audio_coding_dep = declare_dependency( - link_with: libaudio_coding +webrtc_audio_coding_dep = declare_dependency( + link_with: libwebrtc_audio_coding, + include_directories: webrtc_inc, + version: meson.project_version() +) + +install_headers(['codecs/isac/bandwidth_info.h'], + subdir: 'webrtc_audio_processing/webrtc/modules/audio_coding/codecs/isac' +) + +install_headers(['codecs/isac/main/source/settings.h'], + subdir: 'webrtc_audio_processing/webrtc/modules/audio_coding/codecs/isac/main/source' +) + +install_headers(['codecs/isac/main/include/isac.h'], + subdir: 'webrtc_audio_processing/webrtc/modules/audio_coding/codecs/isac/main/include' ) diff --git a/webrtc/modules/audio_processing/meson.build b/webrtc/modules/audio_processing/meson.build index 5ce87a1..34e23e0 100644 --- a/webrtc/modules/audio_processing/meson.build +++ b/webrtc/modules/audio_processing/meson.build @@ -109,7 +109,7 @@ install_headers(webrtc_audio_processing_include_headers, libwebrtc_audio_processing = library('webrtc_audio_processing', webrtc_audio_processing_sources, - dependencies: [base_dep, audio_coding_dep, system_wrappers_dep, common_audio_dep, audio_coding_dep, webrtc_dep] + common_deps, + dependencies: [base_dep, webrtc_audio_coding_dep, system_wrappers_dep, common_audio_dep, webrtc_dep] + common_deps, link_with: extra_libs, include_directories: webrtc_inc, c_args: common_cflags,