AnotherFoxGuy/webrtc-audio-processing
1
0
Fork 0
Code Pull Requests Activity

Bump to WebRTC M120 release

Browse Source 
Some API deprecation -- ExperimentalAgc and ExperimentalNs are gone.
We're continuing to carry iSAC even though it's gone upstream, but maybe
we'll want to drop that soon.
This commit is contained in:
Arun Raghavan
2023-12-12 10:42:58 -05:00
parent 9a202fb8c2
commit c6abf6cd3f
479 changed files with 20900 additions and 11996 deletions
Download Patch File Download Diff File Expand all files Collapse all files

78
webrtc/modules/audio_processing/agc2/agc2_common.h
View File

@ -11,74 +11,50 @@
#ifndef MODULES_AUDIO_PROCESSING_AGC2_AGC2_COMMON_H_
#define MODULES_AUDIO_PROCESSING_AGC2_AGC2_COMMON_H_
#include <stddef.h>
namespace webrtc {
constexpr float kMinFloatS16Value = -32768.f;
constexpr float kMaxFloatS16Value = 32767.f;
constexpr float kMinFloatS16Value = -32768.0f;
constexpr float kMaxFloatS16Value = 32767.0f;
constexpr float kMaxAbsFloatS16Value = 32768.0f;
constexpr size_t kFrameDurationMs = 10;
constexpr size_t kSubFramesInFrame = 20;
constexpr size_t kMaximalNumberOfSamplesPerChannel = 480;
// Minimum audio level in dBFS scale for S16 samples.
constexpr float kMinLevelDbfs = -90.31f;
constexpr float kAttackFilterConstant = 0.f;
constexpr int kFrameDurationMs = 10;
constexpr int kSubFramesInFrame = 20;
constexpr int kMaximalNumberOfSamplesPerChannel = 480;
// Adaptive digital gain applier settings.
// Adaptive digital gain applier settings below.
constexpr float kHeadroomDbfs = 1.f;
constexpr float kMaxGainDb = 30.f;
constexpr float kInitialAdaptiveDigitalGainDb = 8.f;
// At what limiter levels should we start decreasing the adaptive digital gain.
constexpr float kLimiterThresholdForAgcGainDbfs = -kHeadroomDbfs;
constexpr float kLimiterThresholdForAgcGainDbfs = -1.0f;
// This is the threshold for speech. Speech frames are used for updating the
// speech level, measuring the amount of speech, and decide when to allow target
// gain reduction.
constexpr float kVadConfidenceThreshold = 0.9f;
// Number of milliseconds to wait to periodically reset the VAD.
constexpr int kVadResetPeriodMs = 1500;
// The amount of 'memory' of the Level Estimator. Decides leak factors.
constexpr size_t kFullBufferSizeMs = 1200;
constexpr float kFullBufferLeakFactor = 1.f - 1.f / kFullBufferSizeMs;
// Speech probability threshold to detect speech activity.
constexpr float kVadConfidenceThreshold = 0.95f;
constexpr float kInitialSpeechLevelEstimateDbfs = -30.f;
// Minimum number of adjacent speech frames having a sufficiently high speech
// probability to reliably detect speech activity.
constexpr int kAdjacentSpeechFramesThreshold = 12;
// Robust VAD probability and speech decisions.
constexpr float kDefaultSmoothedVadProbabilityAttack = 1.f;
constexpr int kDefaultLevelEstimatorAdjacentSpeechFramesThreshold = 1;
// Number of milliseconds of speech frames to observe to make the estimator
// confident.
constexpr float kLevelEstimatorTimeToConfidenceMs = 400;
constexpr float kLevelEstimatorLeakFactor =
1.0f - 1.0f / kLevelEstimatorTimeToConfidenceMs;
// Saturation Protector settings.
constexpr float kDefaultInitialSaturationMarginDb = 20.f;
constexpr float kDefaultExtraSaturationMarginDb = 2.f;
constexpr size_t kPeakEnveloperSuperFrameLengthMs = 400;
static_assert(kFullBufferSizeMs % kPeakEnveloperSuperFrameLengthMs == 0,
"Full buffer size should be a multiple of super frame length for "
"optimal Saturation Protector performance.");
constexpr size_t kPeakEnveloperBufferSize =
kFullBufferSizeMs / kPeakEnveloperSuperFrameLengthMs + 1;
// This value is 10 ** (-1/20 * frame_size_ms / satproc_attack_ms),
// where satproc_attack_ms is 5000.
constexpr float kSaturationProtectorAttackConstant = 0.9988493699365052f;
// This value is 10 ** (-1/20 * frame_size_ms / satproc_decay_ms),
// where satproc_decay_ms is 1000.
constexpr float kSaturationProtectorDecayConstant = 0.9997697679981565f;
// This is computed from kDecayMs by
// 10 ** (-1/20 * subframe_duration / kDecayMs).
// |subframe_duration| is |kFrameDurationMs / kSubFramesInFrame|.
// kDecayMs is defined in agc2_testing_common.h
constexpr float kDecayFilterConstant = 0.9998848773724686f;
constexpr float kSaturationProtectorInitialHeadroomDb = 20.0f;
constexpr int kSaturationProtectorBufferSize = 4;
// Number of interpolation points for each region of the limiter.
// These values have been tuned to limit the interpolated gain curve error given
// the limiter parameters and allowing a maximum error of +/- 32768^-1.
constexpr size_t kInterpolatedGainCurveKneePoints = 22;
constexpr size_t kInterpolatedGainCurveBeyondKneePoints = 10;
constexpr size_t kInterpolatedGainCurveTotalPoints =
constexpr int kInterpolatedGainCurveKneePoints = 22;
constexpr int kInterpolatedGainCurveBeyondKneePoints = 10;
constexpr int kInterpolatedGainCurveTotalPoints =
kInterpolatedGainCurveKneePoints + kInterpolatedGainCurveBeyondKneePoints;
} // namespace webrtc