| OLD | NEW |
| 1 /* | 1 /* |
| 2 * Copyright (c) 2016 The WebRTC project authors. All Rights Reserved. | 2 * Copyright (c) 2016 The WebRTC project authors. All Rights Reserved. |
| 3 * | 3 * |
| 4 * Use of this source code is governed by a BSD-style license | 4 * Use of this source code is governed by a BSD-style license |
| 5 * that can be found in the LICENSE file in the root of the source | 5 * that can be found in the LICENSE file in the root of the source |
| 6 * tree. An additional intellectual property rights grant can be found | 6 * tree. An additional intellectual property rights grant can be found |
| 7 * in the file PATENTS. All contributing project authors may | 7 * in the file PATENTS. All contributing project authors may |
| 8 * be found in the AUTHORS file in the root of the source tree. | 8 * be found in the AUTHORS file in the root of the source tree. |
| 9 */ | 9 */ |
| 10 | 10 |
| 11 #include "webrtc/modules/audio_processing/residual_echo_detector.h" | 11 #include "webrtc/modules/audio_processing/residual_echo_detector.h" |
| 12 | 12 |
| 13 #include <algorithm> | 13 #include <algorithm> |
| 14 #include <numeric> | 14 #include <numeric> |
| 15 | 15 |
| 16 #include "webrtc/base/atomicops.h" |
| 16 #include "webrtc/modules/audio_processing/audio_buffer.h" | 17 #include "webrtc/modules/audio_processing/audio_buffer.h" |
| 18 #include "webrtc/modules/audio_processing/logging/apm_data_dumper.h" |
| 17 #include "webrtc/system_wrappers/include/metrics.h" | 19 #include "webrtc/system_wrappers/include/metrics.h" |
| 18 | 20 |
| 19 namespace { | 21 namespace { |
| 20 | 22 |
| 21 float Power(rtc::ArrayView<const float> input) { | 23 float Power(rtc::ArrayView<const float> input) { |
| 22 return std::inner_product(input.begin(), input.end(), input.begin(), 0.f); | 24 if (input.size() == 0) { |
| 25 return 0.f; |
| 26 } |
| 27 return std::inner_product(input.begin(), input.end(), input.begin(), 0.f) / |
| 28 input.size(); |
| 23 } | 29 } |
| 24 | 30 |
| 25 constexpr size_t kLookbackFrames = 650; | 31 constexpr size_t kLookbackFrames = 650; |
| 26 // TODO(ivoc): Verify the size of this buffer. | 32 // TODO(ivoc): Verify the size of this buffer. |
| 27 constexpr size_t kRenderBufferSize = 30; | 33 constexpr size_t kRenderBufferSize = 30; |
| 28 constexpr float kAlpha = 0.001f; | 34 constexpr float kAlpha = 0.001f; |
| 29 // 10 seconds of data, updated every 10 ms. | 35 // 10 seconds of data, updated every 10 ms. |
| 30 constexpr size_t kAggregationBufferSize = 10 * 100; | 36 constexpr size_t kAggregationBufferSize = 10 * 100; |
| 31 | 37 |
| 32 } // namespace | 38 } // namespace |
| 33 | 39 |
| 34 namespace webrtc { | 40 namespace webrtc { |
| 35 | 41 |
| 42 int ResidualEchoDetector::instance_count_ = 0; |
| 43 |
| 36 ResidualEchoDetector::ResidualEchoDetector() | 44 ResidualEchoDetector::ResidualEchoDetector() |
| 37 : render_buffer_(kRenderBufferSize), | 45 : data_dumper_( |
| 46 new ApmDataDumper(rtc::AtomicOps::Increment(&instance_count_))), |
| 47 render_buffer_(kRenderBufferSize), |
| 38 render_power_(kLookbackFrames), | 48 render_power_(kLookbackFrames), |
| 39 render_power_mean_(kLookbackFrames), | 49 render_power_mean_(kLookbackFrames), |
| 40 render_power_std_dev_(kLookbackFrames), | 50 render_power_std_dev_(kLookbackFrames), |
| 41 covariances_(kLookbackFrames), | 51 covariances_(kLookbackFrames), |
| 42 recent_likelihood_max_(kAggregationBufferSize) {} | 52 recent_likelihood_max_(kAggregationBufferSize) {} |
| 43 | 53 |
| 44 ResidualEchoDetector::~ResidualEchoDetector() = default; | 54 ResidualEchoDetector::~ResidualEchoDetector() = default; |
| 45 | 55 |
| 46 void ResidualEchoDetector::AnalyzeRenderAudio( | 56 void ResidualEchoDetector::AnalyzeRenderAudio( |
| 47 rtc::ArrayView<const float> render_audio) { | 57 rtc::ArrayView<const float> render_audio) { |
| 58 // Dump debug data assuming 48 kHz sample rate (if this assumption is not |
| 59 // valid the dumped audio will need to be converted offline accordingly). |
| 60 data_dumper_->DumpWav("ed_render", render_audio.size(), render_audio.data(), |
| 61 48000, 1); |
| 62 |
| 48 if (render_buffer_.Size() == 0) { | 63 if (render_buffer_.Size() == 0) { |
| 49 frames_since_zero_buffer_size_ = 0; | 64 frames_since_zero_buffer_size_ = 0; |
| 50 } else if (frames_since_zero_buffer_size_ >= kRenderBufferSize) { | 65 } else if (frames_since_zero_buffer_size_ >= kRenderBufferSize) { |
| 51 // This can happen in a few cases: at the start of a call, due to a glitch | 66 // This can happen in a few cases: at the start of a call, due to a glitch |
| 52 // or due to clock drift. The excess capture value will be ignored. | 67 // or due to clock drift. The excess capture value will be ignored. |
| 53 // TODO(ivoc): Include how often this happens in APM stats. | 68 // TODO(ivoc): Include how often this happens in APM stats. |
| 54 render_buffer_.Pop(); | 69 render_buffer_.Pop(); |
| 55 frames_since_zero_buffer_size_ = 0; | 70 frames_since_zero_buffer_size_ = 0; |
| 56 } | 71 } |
| 57 ++frames_since_zero_buffer_size_; | 72 ++frames_since_zero_buffer_size_; |
| 58 float power = Power(render_audio); | 73 float power = Power(render_audio); |
| 59 render_buffer_.Push(power); | 74 render_buffer_.Push(power); |
| 60 } | 75 } |
| 61 | 76 |
| 62 void ResidualEchoDetector::AnalyzeCaptureAudio( | 77 void ResidualEchoDetector::AnalyzeCaptureAudio( |
| 63 rtc::ArrayView<const float> capture_audio) { | 78 rtc::ArrayView<const float> capture_audio) { |
| 79 // Dump debug data assuming 48 kHz sample rate (if this assumption is not |
| 80 // valid the dumped audio will need to be converted offline accordingly). |
| 81 data_dumper_->DumpWav("ed_capture", capture_audio.size(), |
| 82 capture_audio.data(), 48000, 1); |
| 83 |
| 64 if (first_process_call_) { | 84 if (first_process_call_) { |
| 65 // On the first process call (so the start of a call), we must flush the | 85 // On the first process call (so the start of a call), we must flush the |
| 66 // render buffer, otherwise the render data will be delayed. | 86 // render buffer, otherwise the render data will be delayed. |
| 67 render_buffer_.Clear(); | 87 render_buffer_.Clear(); |
| 68 first_process_call_ = false; | 88 first_process_call_ = false; |
| 69 } | 89 } |
| 70 | 90 |
| 71 // Get the next render value. | 91 // Get the next render value. |
| 72 const rtc::Optional<float> buffered_render_power = render_buffer_.Pop(); | 92 const rtc::Optional<float> buffered_render_power = render_buffer_.Pop(); |
| 73 if (!buffered_render_power) { | 93 if (!buffered_render_power) { |
| (...skipping 59 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 133 cov.Clear(); | 153 cov.Clear(); |
| 134 } | 154 } |
| 135 echo_likelihood_ = 0.f; | 155 echo_likelihood_ = 0.f; |
| 136 next_insertion_index_ = 0; | 156 next_insertion_index_ = 0; |
| 137 reliability_ = 0.f; | 157 reliability_ = 0.f; |
| 138 } | 158 } |
| 139 | 159 |
| 140 void ResidualEchoDetector::PackRenderAudioBuffer( | 160 void ResidualEchoDetector::PackRenderAudioBuffer( |
| 141 AudioBuffer* audio, | 161 AudioBuffer* audio, |
| 142 std::vector<float>* packed_buffer) { | 162 std::vector<float>* packed_buffer) { |
| 143 RTC_DCHECK_GE(160, audio->num_frames_per_band()); | |
| 144 | |
| 145 packed_buffer->clear(); | 163 packed_buffer->clear(); |
| 146 packed_buffer->insert(packed_buffer->end(), | 164 packed_buffer->insert(packed_buffer->end(), audio->channels_f()[0], |
| 147 audio->split_bands_const_f(0)[kBand0To8kHz], | 165 audio->channels_f()[0] + audio->num_frames()); |
| 148 (audio->split_bands_const_f(0)[kBand0To8kHz] + | |
| 149 audio->num_frames_per_band())); | |
| 150 } | 166 } |
| 151 | 167 |
| 152 } // namespace webrtc | 168 } // namespace webrtc |
| OLD | NEW |