3e7b7b154b
This CL introduces two related changes 1) It changes the way that the AEC3 determines whether the linear filter is sufficiently good for its output to be used. The new scheme achieves this much earlier than what was done in the legacy scheme. 2) It changes the way that saturated echo is and handled so that the impact of the nearend speech is lower. Bug: webrtc:9835,webrtc:9843,chromium:895435,chromium:895431 Change-Id: I0b493676886e2134205e9992bbe4badac7e414cc Reviewed-on: https://webrtc-review.googlesource.com/c/104380 Commit-Queue: Per Åhgren <peah@webrtc.org> Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org> Cr-Commit-Position: refs/heads/master@{#25208}
327 lines
11 KiB
C++
327 lines
11 KiB
C++
|
|
/*
|
|
* Copyright (c) 2017 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/residual_echo_estimator.h"
|
|
|
|
#include <numeric>
|
|
#include <vector>
|
|
|
|
#include "modules/audio_processing/aec3/reverb_model.h"
|
|
#include "modules/audio_processing/aec3/reverb_model_fallback.h"
|
|
#include "rtc_base/checks.h"
|
|
#include "system_wrappers/include/field_trial.h"
|
|
|
|
namespace webrtc {
|
|
namespace {
|
|
|
|
bool EnableSoftTransparentMode() {
|
|
return !field_trial::IsEnabled("WebRTC-Aec3SoftTransparentModeKillSwitch");
|
|
}
|
|
|
|
bool OverrideEstimatedEchoPathGain() {
|
|
return !field_trial::IsEnabled("WebRTC-Aec3OverrideEchoPathGainKillSwitch");
|
|
}
|
|
|
|
bool UseFixedNonLinearReverbModel() {
|
|
return field_trial::IsEnabled(
|
|
"WebRTC-Aec3StandardNonlinearReverbModelKillSwitch");
|
|
}
|
|
|
|
// Computes the indexes that will be used for computing spectral power over
|
|
// the blocks surrounding the delay.
|
|
void GetRenderIndexesToAnalyze(
|
|
const VectorBuffer& spectrum_buffer,
|
|
const EchoCanceller3Config::EchoModel& echo_model,
|
|
int filter_delay_blocks,
|
|
bool gain_limiter_running,
|
|
int headroom,
|
|
int* idx_start,
|
|
int* idx_stop) {
|
|
RTC_DCHECK(idx_start);
|
|
RTC_DCHECK(idx_stop);
|
|
if (gain_limiter_running) {
|
|
if (static_cast<size_t>(headroom) >
|
|
echo_model.render_post_window_size_init) {
|
|
*idx_start = spectrum_buffer.OffsetIndex(
|
|
spectrum_buffer.read,
|
|
-static_cast<int>(echo_model.render_post_window_size_init));
|
|
} else {
|
|
*idx_start = spectrum_buffer.IncIndex(spectrum_buffer.write);
|
|
}
|
|
|
|
*idx_stop = spectrum_buffer.OffsetIndex(
|
|
spectrum_buffer.read, echo_model.render_pre_window_size_init);
|
|
} else {
|
|
size_t window_start;
|
|
size_t window_end;
|
|
window_start =
|
|
std::max(0, filter_delay_blocks -
|
|
static_cast<int>(echo_model.render_pre_window_size));
|
|
window_end = filter_delay_blocks +
|
|
static_cast<int>(echo_model.render_post_window_size);
|
|
*idx_start =
|
|
spectrum_buffer.OffsetIndex(spectrum_buffer.read, window_start);
|
|
*idx_stop =
|
|
spectrum_buffer.OffsetIndex(spectrum_buffer.read, window_end + 1);
|
|
}
|
|
}
|
|
|
|
} // namespace
|
|
|
|
ResidualEchoEstimator::ResidualEchoEstimator(const EchoCanceller3Config& config)
|
|
: config_(config),
|
|
soft_transparent_mode_(EnableSoftTransparentMode()),
|
|
override_estimated_echo_path_gain_(OverrideEstimatedEchoPathGain()),
|
|
use_fixed_nonlinear_reverb_model_(UseFixedNonLinearReverbModel()) {
|
|
if (config_.ep_strength.reverb_based_on_render) {
|
|
echo_reverb_.reset(new ReverbModel());
|
|
} else {
|
|
echo_reverb_fallback.reset(
|
|
new ReverbModelFallback(config_.filter.main.length_blocks));
|
|
}
|
|
Reset();
|
|
}
|
|
|
|
ResidualEchoEstimator::~ResidualEchoEstimator() = default;
|
|
|
|
void ResidualEchoEstimator::Estimate(
|
|
const AecState& aec_state,
|
|
const RenderBuffer& render_buffer,
|
|
const std::array<float, kFftLengthBy2Plus1>& S2_linear,
|
|
const std::array<float, kFftLengthBy2Plus1>& Y2,
|
|
std::array<float, kFftLengthBy2Plus1>* R2) {
|
|
RTC_DCHECK(R2);
|
|
|
|
// Estimate the power of the stationary noise in the render signal.
|
|
RenderNoisePower(render_buffer, &X2_noise_floor_, &X2_noise_floor_counter_);
|
|
|
|
// Estimate the residual echo power.
|
|
if (aec_state.UsableLinearEstimate()) {
|
|
LinearEstimate(S2_linear, aec_state.Erle(), aec_state.ErleUncertainty(),
|
|
R2);
|
|
|
|
// When there is saturated echo, assume the same spectral content as is
|
|
// present in the micropone signal.
|
|
if (aec_state.SaturatedEcho()) {
|
|
std::copy(Y2.begin(), Y2.end(), R2->begin());
|
|
}
|
|
|
|
// Adds the estimated unmodelled echo power to the residual echo power
|
|
// estimate.
|
|
if (echo_reverb_) {
|
|
echo_reverb_->AddReverb(
|
|
render_buffer.Spectrum(aec_state.FilterLengthBlocks() + 1),
|
|
aec_state.GetReverbFrequencyResponse(), aec_state.ReverbDecay(), *R2);
|
|
|
|
} else {
|
|
RTC_DCHECK(echo_reverb_fallback);
|
|
echo_reverb_fallback->AddEchoReverb(S2_linear,
|
|
aec_state.FilterDelayBlocks(),
|
|
aec_state.ReverbDecay(), R2);
|
|
}
|
|
|
|
} else {
|
|
// Estimate the echo generating signal power.
|
|
std::array<float, kFftLengthBy2Plus1> X2;
|
|
|
|
EchoGeneratingPower(render_buffer.GetSpectrumBuffer(), config_.echo_model,
|
|
render_buffer.Headroom(), aec_state.FilterDelayBlocks(),
|
|
aec_state.IsSuppressionGainLimitActive(),
|
|
!aec_state.UseStationaryProperties(), &X2);
|
|
|
|
// Subtract the stationary noise power to avoid stationary noise causing
|
|
// excessive echo suppression.
|
|
std::transform(X2.begin(), X2.end(), X2_noise_floor_.begin(), X2.begin(),
|
|
[&](float a, float b) {
|
|
return std::max(
|
|
0.f, a - config_.echo_model.stationary_gate_slope * b);
|
|
});
|
|
|
|
float echo_path_gain;
|
|
if (override_estimated_echo_path_gain_) {
|
|
echo_path_gain = aec_state.TransparentMode() && soft_transparent_mode_
|
|
? 0.01f
|
|
: config_.ep_strength.lf;
|
|
} else {
|
|
echo_path_gain = aec_state.TransparentMode() && soft_transparent_mode_
|
|
? 0.01f
|
|
: aec_state.EchoPathGain();
|
|
}
|
|
NonLinearEstimate(echo_path_gain, X2, Y2, R2);
|
|
|
|
// When there is saturated echo, assume the same spectral content as is
|
|
// present in the micropone signal.
|
|
if (aec_state.SaturatedEcho()) {
|
|
std::copy(Y2.begin(), Y2.end(), R2->begin());
|
|
}
|
|
|
|
if (!(aec_state.TransparentMode() && soft_transparent_mode_)) {
|
|
if (echo_reverb_) {
|
|
echo_reverb_->AddReverbNoFreqShaping(
|
|
render_buffer.Spectrum(aec_state.FilterDelayBlocks() + 1),
|
|
echo_path_gain * echo_path_gain, aec_state.ReverbDecay(), *R2);
|
|
} else {
|
|
RTC_DCHECK(echo_reverb_fallback);
|
|
echo_reverb_fallback->AddEchoReverb(*R2,
|
|
config_.filter.main.length_blocks,
|
|
aec_state.ReverbDecay(), R2);
|
|
}
|
|
}
|
|
}
|
|
|
|
if (aec_state.UseStationaryProperties()) {
|
|
// Scale the echo according to echo audibility.
|
|
std::array<float, kFftLengthBy2Plus1> residual_scaling;
|
|
aec_state.GetResidualEchoScaling(residual_scaling);
|
|
for (size_t k = 0; k < R2->size(); ++k) {
|
|
(*R2)[k] *= residual_scaling[k];
|
|
if (residual_scaling[k] == 0.f) {
|
|
R2_hold_counter_[k] = 0;
|
|
}
|
|
}
|
|
}
|
|
if (!soft_transparent_mode_) {
|
|
// If the echo is deemed inaudible, set the residual echo to zero.
|
|
if (aec_state.TransparentMode()) {
|
|
R2->fill(0.f);
|
|
R2_old_.fill(0.f);
|
|
R2_hold_counter_.fill(0.f);
|
|
}
|
|
}
|
|
|
|
std::copy(R2->begin(), R2->end(), R2_old_.begin());
|
|
}
|
|
|
|
void ResidualEchoEstimator::Reset() {
|
|
if (echo_reverb_) {
|
|
echo_reverb_->Reset();
|
|
} else {
|
|
RTC_DCHECK(echo_reverb_fallback);
|
|
echo_reverb_fallback->Reset();
|
|
}
|
|
X2_noise_floor_counter_.fill(config_.echo_model.noise_floor_hold);
|
|
X2_noise_floor_.fill(config_.echo_model.min_noise_floor_power);
|
|
R2_old_.fill(0.f);
|
|
R2_hold_counter_.fill(0.f);
|
|
}
|
|
|
|
void ResidualEchoEstimator::LinearEstimate(
|
|
const std::array<float, kFftLengthBy2Plus1>& S2_linear,
|
|
const std::array<float, kFftLengthBy2Plus1>& erle,
|
|
absl::optional<float> erle_uncertainty,
|
|
std::array<float, kFftLengthBy2Plus1>* R2) {
|
|
std::fill(R2_hold_counter_.begin(), R2_hold_counter_.end(), 10.f);
|
|
if (erle_uncertainty) {
|
|
for (size_t k = 0; k < R2->size(); ++k) {
|
|
(*R2)[k] = S2_linear[k] * *erle_uncertainty;
|
|
}
|
|
} else {
|
|
std::transform(erle.begin(), erle.end(), S2_linear.begin(), R2->begin(),
|
|
[](float a, float b) {
|
|
RTC_DCHECK_LT(0.f, a);
|
|
return b / a;
|
|
});
|
|
}
|
|
}
|
|
|
|
void ResidualEchoEstimator::NonLinearEstimate(
|
|
float echo_path_gain,
|
|
const std::array<float, kFftLengthBy2Plus1>& X2,
|
|
const std::array<float, kFftLengthBy2Plus1>& Y2,
|
|
std::array<float, kFftLengthBy2Plus1>* R2) {
|
|
// Compute preliminary residual echo.
|
|
std::transform(X2.begin(), X2.end(), R2->begin(), [echo_path_gain](float a) {
|
|
return a * echo_path_gain * echo_path_gain;
|
|
});
|
|
|
|
if (use_fixed_nonlinear_reverb_model_) {
|
|
for (size_t k = 0; k < R2->size(); ++k) {
|
|
// Update hold counter.
|
|
R2_hold_counter_[k] = R2_old_[k] < (*R2)[k] ? 0 : R2_hold_counter_[k] + 1;
|
|
|
|
// Compute the residual echo by holding a maximum echo powers and an echo
|
|
// fading corresponding to a room with an RT60 value of about 50 ms.
|
|
(*R2)[k] =
|
|
R2_hold_counter_[k] < config_.echo_model.nonlinear_hold
|
|
? std::max((*R2)[k], R2_old_[k])
|
|
: std::min((*R2)[k] +
|
|
R2_old_[k] * config_.echo_model.nonlinear_release,
|
|
Y2[k]);
|
|
}
|
|
}
|
|
}
|
|
|
|
void ResidualEchoEstimator::EchoGeneratingPower(
|
|
const VectorBuffer& spectrum_buffer,
|
|
const EchoCanceller3Config::EchoModel& echo_model,
|
|
int headroom_spectrum_buffer,
|
|
int filter_delay_blocks,
|
|
bool gain_limiter_running,
|
|
bool apply_noise_gating,
|
|
std::array<float, kFftLengthBy2Plus1>* X2) const {
|
|
int idx_stop, idx_start;
|
|
|
|
RTC_DCHECK(X2);
|
|
GetRenderIndexesToAnalyze(spectrum_buffer, config_.echo_model,
|
|
filter_delay_blocks, gain_limiter_running,
|
|
headroom_spectrum_buffer, &idx_start, &idx_stop);
|
|
|
|
X2->fill(0.f);
|
|
for (int k = idx_start; k != idx_stop; k = spectrum_buffer.IncIndex(k)) {
|
|
std::transform(X2->begin(), X2->end(), spectrum_buffer.buffer[k].begin(),
|
|
X2->begin(),
|
|
[](float a, float b) { return std::max(a, b); });
|
|
}
|
|
|
|
if (apply_noise_gating) {
|
|
// Apply soft noise gate.
|
|
std::for_each(X2->begin(), X2->end(), [&](float& a) {
|
|
if (config_.echo_model.noise_gate_power > a) {
|
|
a = std::max(0.f, a - config_.echo_model.noise_gate_slope *
|
|
(config_.echo_model.noise_gate_power - a));
|
|
}
|
|
});
|
|
}
|
|
}
|
|
|
|
void ResidualEchoEstimator::RenderNoisePower(
|
|
const RenderBuffer& render_buffer,
|
|
std::array<float, kFftLengthBy2Plus1>* X2_noise_floor,
|
|
std::array<int, kFftLengthBy2Plus1>* X2_noise_floor_counter) const {
|
|
RTC_DCHECK(X2_noise_floor);
|
|
RTC_DCHECK(X2_noise_floor_counter);
|
|
|
|
const auto render_power = render_buffer.Spectrum(0);
|
|
RTC_DCHECK_EQ(X2_noise_floor->size(), render_power.size());
|
|
RTC_DCHECK_EQ(X2_noise_floor_counter->size(), render_power.size());
|
|
|
|
// Estimate the stationary noise power in a minimum statistics manner.
|
|
for (size_t k = 0; k < render_power.size(); ++k) {
|
|
// Decrease rapidly.
|
|
if (render_power[k] < (*X2_noise_floor)[k]) {
|
|
(*X2_noise_floor)[k] = render_power[k];
|
|
(*X2_noise_floor_counter)[k] = 0;
|
|
} else {
|
|
// Increase in a delayed, leaky manner.
|
|
if ((*X2_noise_floor_counter)[k] >=
|
|
static_cast<int>(config_.echo_model.noise_floor_hold)) {
|
|
(*X2_noise_floor)[k] =
|
|
std::max((*X2_noise_floor)[k] * 1.1f,
|
|
config_.echo_model.min_noise_floor_power);
|
|
} else {
|
|
++(*X2_noise_floor_counter)[k];
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
} // namespace webrtc
|