1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241
|
/*
* Copyright (c) 2018 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 "modules/audio_processing/aec3/stationarity_estimator.h"
#include <algorithm>
#include <array>
#include "api/array_view.h"
#include "modules/audio_processing/aec3/aec3_common.h"
#include "modules/audio_processing/aec3/spectrum_buffer.h"
#include "modules/audio_processing/logging/apm_data_dumper.h"
namespace webrtc {
namespace {
constexpr float kMinNoisePower = 10.f;
constexpr int kHangoverBlocks = kNumBlocksPerSecond / 20;
constexpr int kNBlocksAverageInitPhase = 20;
constexpr int kNBlocksInitialPhase = kNumBlocksPerSecond * 2.;
} // namespace
StationarityEstimator::StationarityEstimator()
: data_dumper_(new ApmDataDumper(instance_count_.fetch_add(1) + 1)) {
Reset();
}
StationarityEstimator::~StationarityEstimator() = default;
void StationarityEstimator::Reset() {
noise_.Reset();
hangovers_.fill(0);
stationarity_flags_.fill(false);
}
// Update just the noise estimator. Usefull until the delay is known
void StationarityEstimator::UpdateNoiseEstimator(
ArrayView<const std::array<float, kFftLengthBy2Plus1>> spectrum) {
noise_.Update(spectrum);
data_dumper_->DumpRaw("aec3_stationarity_noise_spectrum", noise_.Spectrum());
data_dumper_->DumpRaw("aec3_stationarity_is_block_stationary",
IsBlockStationary());
}
void StationarityEstimator::UpdateStationarityFlags(
const SpectrumBuffer& spectrum_buffer,
ArrayView<const float> render_reverb_contribution_spectrum,
int idx_current,
int num_lookahead) {
std::array<int, kWindowLength> indexes;
int num_lookahead_bounded = std::min(num_lookahead, kWindowLength - 1);
int idx = idx_current;
if (num_lookahead_bounded < kWindowLength - 1) {
int num_lookback = (kWindowLength - 1) - num_lookahead_bounded;
idx = spectrum_buffer.OffsetIndex(idx_current, num_lookback);
}
// For estimating the stationarity properties of the current frame, the
// power for each band is accumulated for several consecutive spectra in the
// method EstimateBandStationarity.
// In order to avoid getting the indexes of the spectra for every band with
// its associated overhead, those indexes are stored in an array and then use
// when the estimation is done.
indexes[0] = idx;
for (size_t k = 1; k < indexes.size(); ++k) {
indexes[k] = spectrum_buffer.DecIndex(indexes[k - 1]);
}
RTC_DCHECK_EQ(
spectrum_buffer.DecIndex(indexes[kWindowLength - 1]),
spectrum_buffer.OffsetIndex(idx_current, -(num_lookahead_bounded + 1)));
for (size_t k = 0; k < stationarity_flags_.size(); ++k) {
stationarity_flags_[k] = EstimateBandStationarity(
spectrum_buffer, render_reverb_contribution_spectrum, indexes, k);
}
UpdateHangover();
SmoothStationaryPerFreq();
}
bool StationarityEstimator::IsBlockStationary() const {
float acum_stationarity = 0.f;
RTC_DCHECK_EQ(stationarity_flags_.size(), kFftLengthBy2Plus1);
for (size_t band = 0; band < stationarity_flags_.size(); ++band) {
bool st = IsBandStationary(band);
acum_stationarity += static_cast<float>(st);
}
return ((acum_stationarity * (1.f / kFftLengthBy2Plus1)) > 0.75f);
}
bool StationarityEstimator::EstimateBandStationarity(
const SpectrumBuffer& spectrum_buffer,
ArrayView<const float> average_reverb,
const std::array<int, kWindowLength>& indexes,
size_t band) const {
constexpr float kThrStationarity = 10.f;
float acum_power = 0.f;
const int num_render_channels =
static_cast<int>(spectrum_buffer.buffer[0].size());
const float one_by_num_channels = 1.f / num_render_channels;
for (auto idx : indexes) {
for (int ch = 0; ch < num_render_channels; ++ch) {
acum_power += spectrum_buffer.buffer[idx][ch][band] * one_by_num_channels;
}
}
acum_power += average_reverb[band];
float noise = kWindowLength * GetStationarityPowerBand(band);
RTC_CHECK_LT(0.f, noise);
bool stationary = acum_power < kThrStationarity * noise;
data_dumper_->DumpRaw("aec3_stationarity_long_ratio", acum_power / noise);
return stationary;
}
bool StationarityEstimator::AreAllBandsStationary() {
for (auto b : stationarity_flags_) {
if (!b)
return false;
}
return true;
}
void StationarityEstimator::UpdateHangover() {
bool reduce_hangover = AreAllBandsStationary();
for (size_t k = 0; k < stationarity_flags_.size(); ++k) {
if (!stationarity_flags_[k]) {
hangovers_[k] = kHangoverBlocks;
} else if (reduce_hangover) {
hangovers_[k] = std::max(hangovers_[k] - 1, 0);
}
}
}
void StationarityEstimator::SmoothStationaryPerFreq() {
std::array<bool, kFftLengthBy2Plus1> all_ahead_stationary_smooth;
for (size_t k = 1; k < kFftLengthBy2Plus1 - 1; ++k) {
all_ahead_stationary_smooth[k] = stationarity_flags_[k - 1] &&
stationarity_flags_[k] &&
stationarity_flags_[k + 1];
}
all_ahead_stationary_smooth[0] = all_ahead_stationary_smooth[1];
all_ahead_stationary_smooth[kFftLengthBy2Plus1 - 1] =
all_ahead_stationary_smooth[kFftLengthBy2Plus1 - 2];
stationarity_flags_ = all_ahead_stationary_smooth;
}
std::atomic<int> StationarityEstimator::instance_count_(0);
StationarityEstimator::NoiseSpectrum::NoiseSpectrum() {
Reset();
}
StationarityEstimator::NoiseSpectrum::~NoiseSpectrum() = default;
void StationarityEstimator::NoiseSpectrum::Reset() {
block_counter_ = 0;
noise_spectrum_.fill(kMinNoisePower);
}
void StationarityEstimator::NoiseSpectrum::Update(
ArrayView<const std::array<float, kFftLengthBy2Plus1>> spectrum) {
RTC_DCHECK_LE(1, spectrum[0].size());
const int num_render_channels = static_cast<int>(spectrum.size());
std::array<float, kFftLengthBy2Plus1> avg_spectrum_data;
ArrayView<const float> avg_spectrum;
if (num_render_channels == 1) {
avg_spectrum = spectrum[0];
} else {
// For multiple channels, average the channel spectra before passing to the
// noise spectrum estimator.
avg_spectrum = avg_spectrum_data;
std::copy(spectrum[0].begin(), spectrum[0].end(),
avg_spectrum_data.begin());
for (int ch = 1; ch < num_render_channels; ++ch) {
for (size_t k = 1; k < kFftLengthBy2Plus1; ++k) {
avg_spectrum_data[k] += spectrum[ch][k];
}
}
const float one_by_num_channels = 1.f / num_render_channels;
for (size_t k = 1; k < kFftLengthBy2Plus1; ++k) {
avg_spectrum_data[k] *= one_by_num_channels;
}
}
++block_counter_;
float alpha = GetAlpha();
for (size_t k = 0; k < kFftLengthBy2Plus1; ++k) {
if (block_counter_ <= kNBlocksAverageInitPhase) {
noise_spectrum_[k] += (1.f / kNBlocksAverageInitPhase) * avg_spectrum[k];
} else {
noise_spectrum_[k] =
UpdateBandBySmoothing(avg_spectrum[k], noise_spectrum_[k], alpha);
}
}
}
float StationarityEstimator::NoiseSpectrum::GetAlpha() const {
constexpr float kAlpha = 0.004f;
constexpr float kAlphaInit = 0.04f;
constexpr float kTiltAlpha = (kAlphaInit - kAlpha) / kNBlocksInitialPhase;
if (block_counter_ > (kNBlocksInitialPhase + kNBlocksAverageInitPhase)) {
return kAlpha;
} else {
return kAlphaInit -
kTiltAlpha * (block_counter_ - kNBlocksAverageInitPhase);
}
}
float StationarityEstimator::NoiseSpectrum::UpdateBandBySmoothing(
float power_band,
float power_band_noise,
float alpha) const {
float power_band_noise_updated = power_band_noise;
if (power_band_noise < power_band) {
RTC_DCHECK_GT(power_band, 0.f);
float alpha_inc = alpha * (power_band_noise / power_band);
if (block_counter_ > kNBlocksInitialPhase) {
if (10.f * power_band_noise < power_band) {
alpha_inc *= 0.1f;
}
}
power_band_noise_updated += alpha_inc * (power_band - power_band_noise);
} else {
power_band_noise_updated += alpha * (power_band - power_band_noise);
power_band_noise_updated =
std::max(power_band_noise_updated, kMinNoisePower);
}
return power_band_noise_updated;
}
} // namespace webrtc
|