mirror of
https://github.com/mollyim/webrtc.git
synced 2025-05-15 06:40:43 +01:00
3e83b7fe8d
Added possibility to extract audio_processing VAD annotations in the Quality Assessment tool. Annotations are extracted into compressed Numpy 'annotations.npz' files. Annotations contain information about VAD, speech level, speech probabilities etc. TBR=alessiob@webrtc.org Bug: webrtc:7494 Change-Id: I0e54bb67132ae4e180f89959b8bca3ea7f259458 Reviewed-on: https://webrtc-review.googlesource.com/17840 Commit-Queue: Alex Loiko <aleloi@webrtc.org> Reviewed-by: Alessio Bazzica <alessiob@webrtc.org> Reviewed-by: Alex Loiko <aleloi@webrtc.org> Cr-Commit-Position: refs/heads/master@{#20581}
85 lines
3 KiB
C++
85 lines
3 KiB
C++
/*
|
|
* 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 "modules/audio_processing/vad/voice_activity_detector.h"
|
|
|
|
#include <algorithm>
|
|
|
|
#include "rtc_base/checks.h"
|
|
|
|
namespace webrtc {
|
|
namespace {
|
|
|
|
const size_t kNumChannels = 1;
|
|
|
|
const double kDefaultVoiceValue = 1.0;
|
|
const double kNeutralProbability = 0.5;
|
|
const double kLowProbability = 0.01;
|
|
|
|
} // namespace
|
|
|
|
VoiceActivityDetector::VoiceActivityDetector()
|
|
: last_voice_probability_(kDefaultVoiceValue),
|
|
standalone_vad_(StandaloneVad::Create()) {
|
|
}
|
|
|
|
VoiceActivityDetector::~VoiceActivityDetector() = default;
|
|
|
|
// Because ISAC has a different chunk length, it updates
|
|
// |chunkwise_voice_probabilities_| and |chunkwise_rms_| when there is new data.
|
|
// Otherwise it clears them.
|
|
void VoiceActivityDetector::ProcessChunk(const int16_t* audio,
|
|
size_t length,
|
|
int sample_rate_hz) {
|
|
RTC_DCHECK_EQ(length, sample_rate_hz / 100);
|
|
// Resample to the required rate.
|
|
const int16_t* resampled_ptr = audio;
|
|
if (sample_rate_hz != kSampleRateHz) {
|
|
RTC_CHECK_EQ(
|
|
resampler_.ResetIfNeeded(sample_rate_hz, kSampleRateHz, kNumChannels),
|
|
0);
|
|
resampler_.Push(audio, length, resampled_, kLength10Ms, length);
|
|
resampled_ptr = resampled_;
|
|
}
|
|
RTC_DCHECK_EQ(length, kLength10Ms);
|
|
|
|
// Each chunk needs to be passed into |standalone_vad_|, because internally it
|
|
// buffers the audio and processes it all at once when GetActivity() is
|
|
// called.
|
|
RTC_CHECK_EQ(standalone_vad_->AddAudio(resampled_ptr, length), 0);
|
|
|
|
audio_processing_.ExtractFeatures(resampled_ptr, length, &features_);
|
|
|
|
chunkwise_voice_probabilities_.resize(features_.num_frames);
|
|
chunkwise_rms_.resize(features_.num_frames);
|
|
std::copy(features_.rms, features_.rms + chunkwise_rms_.size(),
|
|
chunkwise_rms_.begin());
|
|
if (features_.num_frames > 0) {
|
|
if (features_.silence) {
|
|
// The other features are invalid, so set the voice probabilities to an
|
|
// arbitrary low value.
|
|
std::fill(chunkwise_voice_probabilities_.begin(),
|
|
chunkwise_voice_probabilities_.end(), kLowProbability);
|
|
} else {
|
|
std::fill(chunkwise_voice_probabilities_.begin(),
|
|
chunkwise_voice_probabilities_.end(), kNeutralProbability);
|
|
RTC_CHECK_GE(
|
|
standalone_vad_->GetActivity(&chunkwise_voice_probabilities_[0],
|
|
chunkwise_voice_probabilities_.size()),
|
|
0);
|
|
RTC_CHECK_GE(pitch_based_vad_.VoicingProbability(
|
|
features_, &chunkwise_voice_probabilities_[0]),
|
|
0);
|
|
}
|
|
last_voice_probability_ = chunkwise_voice_probabilities_.back();
|
|
}
|
|
}
|
|
|
|
} // namespace webrtc
|