Chromium Code Reviews| OLD | NEW |
|---|---|
| 1 // Copyright 2013 The Chromium Authors. All rights reserved. | 1 // Copyright 2013 The Chromium Authors. All rights reserved. |
| 2 // Use of this source code is governed by a BSD-style license that can be | 2 // Use of this source code is governed by a BSD-style license that can be |
| 3 // found in the LICENSE file. | 3 // found in the LICENSE file. |
| 4 | 4 |
| 5 #include "content/renderer/media/media_stream_audio_processor.h" | 5 #include "content/renderer/media/media_stream_audio_processor.h" |
| 6 | 6 |
| 7 #include "base/command_line.h" | 7 #include "base/command_line.h" |
| 8 #include "base/debug/trace_event.h" | 8 #include "base/debug/trace_event.h" |
| 9 #include "base/metrics/field_trial.h" | 9 #include "base/metrics/field_trial.h" |
| 10 #include "base/metrics/histogram.h" | 10 #include "base/metrics/histogram.h" |
| 11 #include "content/public/common/content_switches.h" | 11 #include "content/public/common/content_switches.h" |
| 12 #include "content/renderer/media/media_stream_audio_processor_options.h" | 12 #include "content/renderer/media/media_stream_audio_processor_options.h" |
| 13 #include "content/renderer/media/rtc_media_constraints.h" | 13 #include "content/renderer/media/rtc_media_constraints.h" |
| 14 #include "media/audio/audio_parameters.h" | 14 #include "media/audio/audio_parameters.h" |
| 15 #include "media/base/audio_converter.h" | 15 #include "media/base/audio_converter.h" |
| 16 #include "media/base/audio_fifo.h" | 16 #include "media/base/audio_fifo.h" |
| 17 #include "media/base/channel_layout.h" | 17 #include "media/base/channel_layout.h" |
| 18 #include "third_party/WebKit/public/platform/WebMediaConstraints.h" | 18 #include "third_party/WebKit/public/platform/WebMediaConstraints.h" |
| 19 #include "third_party/libjingle/source/talk/app/webrtc/mediaconstraintsinterface .h" | 19 #include "third_party/libjingle/source/talk/app/webrtc/mediaconstraintsinterface .h" |
| 20 #include "third_party/webrtc/modules/audio_processing/typing_detection.h" | 20 #include "third_party/webrtc/modules/audio_processing/typing_detection.h" |
| 21 | 21 |
| 22 namespace content { | 22 namespace content { |
| 23 | 23 |
| 24 namespace { | 24 namespace { |
| 25 | 25 |
| 26 using webrtc::AudioProcessing; | 26 using webrtc::AudioProcessing; |
| 27 using webrtc::MediaConstraintsInterface; | |
| 28 | 27 |
| 29 #if defined(OS_ANDROID) | 28 #if defined(OS_ANDROID) |
| 30 const int kAudioProcessingSampleRate = 16000; | 29 const int kAudioProcessingSampleRate = 16000; |
| 31 #else | 30 #else |
| 32 const int kAudioProcessingSampleRate = 32000; | 31 const int kAudioProcessingSampleRate = 32000; |
| 33 #endif | 32 #endif |
| 34 const int kAudioProcessingNumberOfChannels = 1; | 33 const int kAudioProcessingNumberOfChannels = 1; |
| 35 | 34 |
| 36 const int kMaxNumberOfBuffersInFifo = 2; | 35 const int kMaxNumberOfBuffersInFifo = 2; |
| 37 | 36 |
| (...skipping 112 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... | |
| 150 // TODO(xians): consider using SincResampler to save some memcpy. | 149 // TODO(xians): consider using SincResampler to save some memcpy. |
| 151 // Handles mixing and resampling between input and output parameters. | 150 // Handles mixing and resampling between input and output parameters. |
| 152 media::AudioConverter audio_converter_; | 151 media::AudioConverter audio_converter_; |
| 153 scoped_ptr<media::AudioBus> audio_wrapper_; | 152 scoped_ptr<media::AudioBus> audio_wrapper_; |
| 154 scoped_ptr<media::AudioFifo> fifo_; | 153 scoped_ptr<media::AudioFifo> fifo_; |
| 155 }; | 154 }; |
| 156 | 155 |
| 157 MediaStreamAudioProcessor::MediaStreamAudioProcessor( | 156 MediaStreamAudioProcessor::MediaStreamAudioProcessor( |
| 158 const blink::WebMediaConstraints& constraints, | 157 const blink::WebMediaConstraints& constraints, |
| 159 int effects, | 158 int effects, |
| 160 MediaStreamType type, | |
| 161 WebRtcPlayoutDataSource* playout_data_source) | 159 WebRtcPlayoutDataSource* playout_data_source) |
| 162 : render_delay_ms_(0), | 160 : render_delay_ms_(0), |
| 163 playout_data_source_(playout_data_source), | 161 playout_data_source_(playout_data_source), |
| 164 audio_mirroring_(false), | 162 goog_audio_mirroring_(false), |
| 165 typing_detected_(false) { | 163 goog_typing_detected_(false) { |
| 166 capture_thread_checker_.DetachFromThread(); | 164 capture_thread_checker_.DetachFromThread(); |
| 167 render_thread_checker_.DetachFromThread(); | 165 render_thread_checker_.DetachFromThread(); |
| 168 InitializeAudioProcessingModule(constraints, effects, type); | 166 InitializeAudioProcessingModule(constraints, effects); |
| 169 } | 167 } |
| 170 | 168 |
| 171 MediaStreamAudioProcessor::~MediaStreamAudioProcessor() { | 169 MediaStreamAudioProcessor::~MediaStreamAudioProcessor() { |
| 172 DCHECK(main_thread_checker_.CalledOnValidThread()); | 170 DCHECK(main_thread_checker_.CalledOnValidThread()); |
| 173 StopAudioProcessing(); | 171 StopAudioProcessing(); |
| 174 } | 172 } |
| 175 | 173 |
| 176 void MediaStreamAudioProcessor::OnCaptureFormatChanged( | 174 void MediaStreamAudioProcessor::OnCaptureFormatChanged( |
| 177 const media::AudioParameters& source_params) { | 175 const media::AudioParameters& source_params) { |
| 178 DCHECK(main_thread_checker_.CalledOnValidThread()); | 176 DCHECK(main_thread_checker_.CalledOnValidThread()); |
| 179 // There is no need to hold a lock here since the caller guarantees that | 177 // There is no need to hold a lock here since the caller guarantees that |
| 180 // there is no more PushCaptureData() and ProcessAndConsumeData() callbacks | 178 // there is no more PushCaptureData() and ProcessAndConsumeData() callbacks |
| 181 // on the capture thread. | 179 // on the capture thread. |
| 182 InitializeCaptureConverter(source_params); | 180 InitializeCaptureConverter(source_params); |
| 183 | 181 |
| 184 // Reset the |capture_thread_checker_| since the capture data will come from | 182 // Reset the |capture_thread_checker_| since the capture data will come from |
| 185 // a new capture thread. | 183 // a new capture thread. |
| 186 capture_thread_checker_.DetachFromThread(); | 184 capture_thread_checker_.DetachFromThread(); |
| 187 } | 185 } |
| 188 | 186 |
| 189 void MediaStreamAudioProcessor::PushCaptureData(media::AudioBus* audio_source) { | 187 void MediaStreamAudioProcessor::PushCaptureData(media::AudioBus* audio_source) { |
| 190 DCHECK(capture_thread_checker_.CalledOnValidThread()); | 188 DCHECK(capture_thread_checker_.CalledOnValidThread()); |
| 191 DCHECK_EQ(audio_source->channels(), | 189 DCHECK_EQ(audio_source->channels(), |
| 192 capture_converter_->source_parameters().channels()); | 190 capture_converter_->source_parameters().channels()); |
| 193 DCHECK_EQ(audio_source->frames(), | 191 DCHECK_EQ(audio_source->frames(), |
| 194 capture_converter_->source_parameters().frames_per_buffer()); | 192 capture_converter_->source_parameters().frames_per_buffer()); |
| 195 | 193 |
| 196 if (audio_mirroring_ && | 194 if (goog_audio_mirroring_ && |
| 197 capture_converter_->source_parameters().channel_layout() == | 195 capture_converter_->source_parameters().channel_layout() == |
| 198 media::CHANNEL_LAYOUT_STEREO) { | 196 media::CHANNEL_LAYOUT_STEREO) { |
| 199 // Swap the first and second channels. | 197 // Swap the first and second channels. |
| 200 audio_source->SwapChannels(0, 1); | 198 audio_source->SwapChannels(0, 1); |
| 201 } | 199 } |
| 202 | 200 |
| 203 capture_converter_->Push(audio_source); | 201 capture_converter_->Push(audio_source); |
| 204 } | 202 } |
| 205 | 203 |
| 206 bool MediaStreamAudioProcessor::ProcessAndConsumeData( | 204 bool MediaStreamAudioProcessor::ProcessAndConsumeData( |
| (...skipping 57 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... | |
| 264 void MediaStreamAudioProcessor::OnPlayoutDataSourceChanged() { | 262 void MediaStreamAudioProcessor::OnPlayoutDataSourceChanged() { |
| 265 DCHECK(main_thread_checker_.CalledOnValidThread()); | 263 DCHECK(main_thread_checker_.CalledOnValidThread()); |
| 266 // There is no need to hold a lock here since the caller guarantees that | 264 // There is no need to hold a lock here since the caller guarantees that |
| 267 // there is no more OnPlayoutData() callback on the render thread. | 265 // there is no more OnPlayoutData() callback on the render thread. |
| 268 render_thread_checker_.DetachFromThread(); | 266 render_thread_checker_.DetachFromThread(); |
| 269 render_converter_.reset(); | 267 render_converter_.reset(); |
| 270 } | 268 } |
| 271 | 269 |
| 272 void MediaStreamAudioProcessor::GetStats(AudioProcessorStats* stats) { | 270 void MediaStreamAudioProcessor::GetStats(AudioProcessorStats* stats) { |
| 273 stats->typing_noise_detected = | 271 stats->typing_noise_detected = |
| 274 (base::subtle::Acquire_Load(&typing_detected_) != false); | 272 (base::subtle::Acquire_Load(&goog_typing_detected_) != false); |
| 275 GetAecStats(audio_processing_.get(), stats); | 273 GetAecStats(audio_processing_.get(), stats); |
| 276 } | 274 } |
| 277 | 275 |
| 278 void MediaStreamAudioProcessor::InitializeAudioProcessingModule( | 276 void MediaStreamAudioProcessor::InitializeAudioProcessingModule( |
| 279 const blink::WebMediaConstraints& constraints, int effects, | 277 const blink::WebMediaConstraints& constraints, int effects) { |
| 280 MediaStreamType type) { | |
| 281 DCHECK(!audio_processing_); | 278 DCHECK(!audio_processing_); |
| 282 | 279 |
| 283 RTCMediaConstraints native_constraints(constraints); | 280 MediaAudioConstraints audio_constraints(constraints, effects); |
| 281 DCHECK(audio_constraints.IsValid()); | |
|
perkj_chrome
2014/04/14 12:15:19
I don't think you should crash if a user user inse
no longer working on chromium
2014/04/14 14:40:50
I was thinking WebRtcAudioCapture::Initialize() wi
| |
| 284 | 282 |
| 285 // Audio mirroring can be enabled even though audio processing is otherwise | 283 // Audio mirroring can be enabled even though audio processing is otherwise |
| 286 // disabled. | 284 // disabled. |
| 287 audio_mirroring_ = GetPropertyFromConstraints( | 285 goog_audio_mirroring_ = audio_constraints.GetProperty( |
| 288 &native_constraints, webrtc::MediaConstraintsInterface::kAudioMirroring); | 286 MediaAudioConstraints::kGoogAudioMirroring); |
| 289 | 287 |
| 290 if (!IsAudioTrackProcessingEnabled()) { | 288 if (!IsAudioTrackProcessingEnabled()) { |
| 291 RecordProcessingState(AUDIO_PROCESSING_IN_WEBRTC); | 289 RecordProcessingState(AUDIO_PROCESSING_IN_WEBRTC); |
| 292 return; | 290 return; |
| 293 } | 291 } |
| 294 | 292 |
| 295 // Only apply the fixed constraints for gUM of MEDIA_DEVICE_AUDIO_CAPTURE. | 293 // |kEchoCancellation| is used as a master control on enabling/disabling |
| 296 DCHECK(IsAudioMediaType(type)); | 294 // the audio processing. |
| 297 if (type == MEDIA_DEVICE_AUDIO_CAPTURE) | 295 const bool echo_cancellation = audio_constraints.GetProperty( |
| 298 ApplyFixedAudioConstraints(&native_constraints); | 296 MediaAudioConstraints::kEchoCancellation); |
| 299 | 297 if (!echo_cancellation) { |
| 300 if (effects & media::AudioParameters::ECHO_CANCELLER) { | 298 RecordProcessingState(AUDIO_PROCESSING_DISABLED); |
| 301 // If platform echo canceller is enabled, disable the software AEC. | 299 return; |
| 302 native_constraints.AddMandatory( | |
| 303 MediaConstraintsInterface::kEchoCancellation, | |
| 304 MediaConstraintsInterface::kValueFalse, true); | |
| 305 } | 300 } |
| 306 | 301 |
| 307 #if defined(OS_IOS) | 302 #if defined(OS_IOS) |
| 308 // On iOS, VPIO provides built-in AEC and AGC. | 303 // On iOS, VPIO provides built-in AEC and AGC. |
| 309 const bool enable_aec = false; | 304 const bool goog_aec = false; |
| 310 const bool enable_agc = false; | 305 const bool goog_agc = false; |
| 311 #else | 306 #else |
| 312 const bool enable_aec = GetPropertyFromConstraints( | 307 // TODO(xians): goog_aec should be just echo_cancellation. |
| 313 &native_constraints, MediaConstraintsInterface::kEchoCancellation); | 308 const bool goog_aec = audio_constraints.GetProperty( |
| 314 const bool enable_agc = GetPropertyFromConstraints( | 309 MediaAudioConstraints::kGoogEchoCancellation); |
| 315 &native_constraints, webrtc::MediaConstraintsInterface::kAutoGainControl); | 310 const bool goog_agc = audio_constraints.GetProperty( |
| 311 MediaAudioConstraints::kGoogAutoGainControl); | |
| 316 #endif | 312 #endif |
| 317 | 313 |
| 318 #if defined(OS_IOS) || defined(OS_ANDROID) | 314 #if defined(OS_IOS) || defined(OS_ANDROID) |
| 319 const bool enable_experimental_aec = false; | 315 const bool goog_experimental_aec = false; |
| 320 const bool enable_typing_detection = false; | 316 const bool goog_typing_detection = false; |
| 321 #else | 317 #else |
| 322 const bool enable_experimental_aec = GetPropertyFromConstraints( | 318 const bool goog_experimental_aec = audio_constraints.GetProperty( |
| 323 &native_constraints, | 319 MediaAudioConstraints::kGoogExperimentalEchoCancellation); |
| 324 MediaConstraintsInterface::kExperimentalEchoCancellation); | 320 const bool goog_typing_detection = audio_constraints.GetProperty( |
| 325 const bool enable_typing_detection = GetPropertyFromConstraints( | 321 MediaAudioConstraints::kGoogTypingNoiseDetection); |
| 326 &native_constraints, MediaConstraintsInterface::kTypingNoiseDetection); | |
| 327 #endif | 322 #endif |
| 328 | 323 |
| 329 const bool enable_ns = GetPropertyFromConstraints( | 324 const bool goog_ns = audio_constraints.GetProperty( |
| 330 &native_constraints, MediaConstraintsInterface::kNoiseSuppression); | 325 MediaAudioConstraints::kGoogNoiseSuppression); |
| 331 const bool enable_experimental_ns = GetPropertyFromConstraints( | 326 const bool goog_experimental_ns = audio_constraints.GetProperty( |
| 332 &native_constraints, | 327 MediaAudioConstraints::kGoogExperimentalNoiseSuppression); |
| 333 MediaConstraintsInterface::kExperimentalNoiseSuppression); | 328 const bool goog_high_pass_filter = audio_constraints.GetProperty( |
| 334 const bool enable_high_pass_filter = GetPropertyFromConstraints( | 329 MediaAudioConstraints::kGoogHighpassFilter); |
| 335 &native_constraints, MediaConstraintsInterface::kHighpassFilter); | |
| 336 | 330 |
| 337 // Return immediately if no audio processing component is enabled. | 331 // Return immediately if no goog constraint is enabled. |
| 338 if (!enable_aec && !enable_experimental_aec && !enable_ns && | 332 if (!goog_aec && !goog_experimental_aec && !goog_ns && |
| 339 !enable_high_pass_filter && !enable_typing_detection && !enable_agc && | 333 !goog_high_pass_filter && !goog_typing_detection && |
| 340 !enable_experimental_ns) { | 334 !goog_agc && !goog_experimental_ns) { |
| 341 RecordProcessingState(AUDIO_PROCESSING_DISABLED); | 335 RecordProcessingState(AUDIO_PROCESSING_DISABLED); |
| 342 return; | 336 return; |
| 343 } | 337 } |
| 344 | 338 |
| 345 // Create and configure the webrtc::AudioProcessing. | 339 // Create and configure the webrtc::AudioProcessing. |
| 346 audio_processing_.reset(webrtc::AudioProcessing::Create(0)); | 340 audio_processing_.reset(webrtc::AudioProcessing::Create(0)); |
| 347 | 341 |
| 348 // Enable the audio processing components. | 342 // Enable the audio processing components. |
| 349 if (enable_aec) { | 343 if (goog_aec) { |
| 350 EnableEchoCancellation(audio_processing_.get()); | 344 EnableEchoCancellation(audio_processing_.get()); |
| 351 if (enable_experimental_aec) | 345 |
| 346 if (goog_experimental_aec) | |
| 352 EnableExperimentalEchoCancellation(audio_processing_.get()); | 347 EnableExperimentalEchoCancellation(audio_processing_.get()); |
| 353 | 348 |
| 354 if (playout_data_source_) | 349 if (playout_data_source_) |
| 355 playout_data_source_->AddPlayoutSink(this); | 350 playout_data_source_->AddPlayoutSink(this); |
| 356 } | 351 } |
| 357 | 352 |
| 358 if (enable_ns) | 353 if (goog_ns) |
| 359 EnableNoiseSuppression(audio_processing_.get()); | 354 EnableNoiseSuppression(audio_processing_.get()); |
| 360 | 355 |
| 361 if (enable_experimental_ns) | 356 if (goog_experimental_ns) |
| 362 EnableExperimentalNoiseSuppression(audio_processing_.get()); | 357 EnableExperimentalNoiseSuppression(audio_processing_.get()); |
| 363 | 358 |
| 364 if (enable_high_pass_filter) | 359 if (goog_high_pass_filter) |
| 365 EnableHighPassFilter(audio_processing_.get()); | 360 EnableHighPassFilter(audio_processing_.get()); |
| 366 | 361 |
| 367 if (enable_typing_detection) { | 362 if (goog_typing_detection) { |
| 368 // TODO(xians): Remove this |typing_detector_| after the typing suppression | 363 // TODO(xians): Remove this |typing_detector_| after the typing suppression |
| 369 // is enabled by default. | 364 // is enabled by default. |
| 370 typing_detector_.reset(new webrtc::TypingDetection()); | 365 typing_detector_.reset(new webrtc::TypingDetection()); |
| 371 EnableTypingDetection(audio_processing_.get(), typing_detector_.get()); | 366 EnableTypingDetection(audio_processing_.get(), typing_detector_.get()); |
| 372 } | 367 } |
| 373 | 368 |
| 374 if (enable_agc) | 369 if (goog_agc) |
| 375 EnableAutomaticGainControl(audio_processing_.get()); | 370 EnableAutomaticGainControl(audio_processing_.get()); |
| 376 | 371 |
| 377 // Configure the audio format the audio processing is running on. This | 372 // Configure the audio format the audio processing is running on. This |
| 378 // has to be done after all the needed components are enabled. | 373 // has to be done after all the needed components are enabled. |
| 379 CHECK_EQ(0, | 374 CHECK_EQ(0, |
| 380 audio_processing_->set_sample_rate_hz(kAudioProcessingSampleRate)); | 375 audio_processing_->set_sample_rate_hz(kAudioProcessingSampleRate)); |
| 381 CHECK_EQ(0, audio_processing_->set_num_channels( | 376 CHECK_EQ(0, audio_processing_->set_num_channels( |
| 382 kAudioProcessingNumberOfChannels, kAudioProcessingNumberOfChannels)); | 377 kAudioProcessingNumberOfChannels, kAudioProcessingNumberOfChannels)); |
| 383 | 378 |
| 384 RecordProcessingState(AUDIO_PROCESSING_ENABLED); | 379 RecordProcessingState(AUDIO_PROCESSING_ENABLED); |
| (...skipping 96 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... | |
| 481 audio_processing_->set_stream_key_pressed(key_pressed); | 476 audio_processing_->set_stream_key_pressed(key_pressed); |
| 482 | 477 |
| 483 err = audio_processing_->ProcessStream(audio_frame); | 478 err = audio_processing_->ProcessStream(audio_frame); |
| 484 DCHECK_EQ(err, 0) << "ProcessStream() error: " << err; | 479 DCHECK_EQ(err, 0) << "ProcessStream() error: " << err; |
| 485 | 480 |
| 486 if (typing_detector_ && | 481 if (typing_detector_ && |
| 487 audio_frame->vad_activity_ != webrtc::AudioFrame::kVadUnknown) { | 482 audio_frame->vad_activity_ != webrtc::AudioFrame::kVadUnknown) { |
| 488 bool vad_active = | 483 bool vad_active = |
| 489 (audio_frame->vad_activity_ == webrtc::AudioFrame::kVadActive); | 484 (audio_frame->vad_activity_ == webrtc::AudioFrame::kVadActive); |
| 490 bool typing_detected = typing_detector_->Process(key_pressed, vad_active); | 485 bool typing_detected = typing_detector_->Process(key_pressed, vad_active); |
| 491 base::subtle::Release_Store(&typing_detected_, typing_detected); | 486 base::subtle::Release_Store(&goog_typing_detected_, typing_detected); |
| 492 } | 487 } |
| 493 | 488 |
| 494 // Return 0 if the volume has not been changed, otherwise return the new | 489 // Return 0 if the volume has not been changed, otherwise return the new |
| 495 // volume. | 490 // volume. |
| 496 return (agc->stream_analog_level() == volume) ? | 491 return (agc->stream_analog_level() == volume) ? |
| 497 0 : agc->stream_analog_level(); | 492 0 : agc->stream_analog_level(); |
| 498 } | 493 } |
| 499 | 494 |
| 500 void MediaStreamAudioProcessor::StopAudioProcessing() { | 495 void MediaStreamAudioProcessor::StopAudioProcessing() { |
| 501 if (!audio_processing_.get()) | 496 if (!audio_processing_.get()) |
| 502 return; | 497 return; |
| 503 | 498 |
| 504 StopAecDump(); | 499 StopAecDump(); |
| 505 | 500 |
| 506 if (playout_data_source_) | 501 if (playout_data_source_) |
| 507 playout_data_source_->RemovePlayoutSink(this); | 502 playout_data_source_->RemovePlayoutSink(this); |
| 508 | 503 |
| 509 audio_processing_.reset(); | 504 audio_processing_.reset(); |
| 510 } | 505 } |
| 511 | 506 |
| 512 bool MediaStreamAudioProcessor::IsAudioTrackProcessingEnabled() const { | 507 bool MediaStreamAudioProcessor::IsAudioTrackProcessingEnabled() const { |
| 513 const std::string group_name = | 508 const std::string group_name = |
| 514 base::FieldTrialList::FindFullName("MediaStreamAudioTrackProcessing"); | 509 base::FieldTrialList::FindFullName("MediaStreamAudioTrackProcessing"); |
| 515 return group_name == "Enabled" || CommandLine::ForCurrentProcess()->HasSwitch( | 510 return group_name == "Enabled" || CommandLine::ForCurrentProcess()->HasSwitch( |
| 516 switches::kEnableAudioTrackProcessing); | 511 switches::kEnableAudioTrackProcessing); |
| 517 } | 512 } |
| 518 | 513 |
| 519 } // namespace content | 514 } // namespace content |
| OLD | NEW |