Add empty residual echo detector.

webrtc/modules/audio_processing/audio_processing_impl.cc

 /*
  *  Copyright (c) 2012 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.
  */
 
@@ skipping 17 matching lines @@
 #include "webrtc/modules/audio_processing/echo_control_mobile_impl.h"
 #include "webrtc/modules/audio_processing/gain_control_for_experimental_agc.h"
 #include "webrtc/modules/audio_processing/gain_control_impl.h"
 #include "webrtc/modules/audio_processing/high_pass_filter_impl.h"
 #if WEBRTC_INTELLIGIBILITY_ENHANCER
 #include "webrtc/modules/audio_processing/intelligibility/intelligibility_enhancer.h"
 #endif
 #include "webrtc/modules/audio_processing/level_controller/level_controller.h"
 #include "webrtc/modules/audio_processing/level_estimator_impl.h"
 #include "webrtc/modules/audio_processing/noise_suppression_impl.h"
+#include "webrtc/modules/audio_processing/residual_echo_detector.h"
 #include "webrtc/modules/audio_processing/transient/transient_suppressor.h"
 #include "webrtc/modules/audio_processing/voice_detection_impl.h"
 #include "webrtc/modules/include/module_common_types.h"
 #include "webrtc/system_wrappers/include/file_wrapper.h"
 #include "webrtc/system_wrappers/include/logging.h"
 #include "webrtc/system_wrappers/include/metrics.h"
 
 #ifdef WEBRTC_AUDIOPROC_DEBUG_DUMP
 // Files generated at build-time by the protobuf compiler.
 #ifdef WEBRTC_ANDROID_PLATFORM_BUILD
@@ skipping 82 matching lines @@
 
 // Throughout webrtc, it's assumed that success is represented by zero.
 static_assert(AudioProcessing::kNoError == 0, "kNoError must be zero");
 
 AudioProcessingImpl::ApmSubmoduleStates::ApmSubmoduleStates() {}
 
 bool AudioProcessingImpl::ApmSubmoduleStates::Update(
     bool high_pass_filter_enabled,
     bool echo_canceller_enabled,
     bool mobile_echo_controller_enabled,
+    bool residual_echo_detector_enabled,
     bool noise_suppressor_enabled,
     bool intelligibility_enhancer_enabled,
     bool beamformer_enabled,
     bool adaptive_gain_controller_enabled,
     bool level_controller_enabled,
     bool voice_activity_detector_enabled,
     bool level_estimator_enabled,
     bool transient_suppressor_enabled) {
   bool changed = false;
   changed |= (high_pass_filter_enabled != high_pass_filter_enabled_);
   changed |= (echo_canceller_enabled != echo_canceller_enabled_);
   changed |=
       (mobile_echo_controller_enabled != mobile_echo_controller_enabled_);
+  changed |=
+      (residual_echo_detector_enabled != residual_echo_detector_enabled_);
   changed |= (noise_suppressor_enabled != noise_suppressor_enabled_);
   changed |=
       (intelligibility_enhancer_enabled != intelligibility_enhancer_enabled_);
   changed |= (beamformer_enabled != beamformer_enabled_);
   changed |=
       (adaptive_gain_controller_enabled != adaptive_gain_controller_enabled_);
   changed |= (level_controller_enabled != level_controller_enabled_);
   changed |= (level_estimator_enabled != level_estimator_enabled_);
   changed |=
       (voice_activity_detector_enabled != voice_activity_detector_enabled_);
   changed |= (transient_suppressor_enabled != transient_suppressor_enabled_);
   if (changed) {
     high_pass_filter_enabled_ = high_pass_filter_enabled;
     echo_canceller_enabled_ = echo_canceller_enabled;
     mobile_echo_controller_enabled_ = mobile_echo_controller_enabled;
+    residual_echo_detector_enabled_ = residual_echo_detector_enabled;
     noise_suppressor_enabled_ = noise_suppressor_enabled;
     intelligibility_enhancer_enabled_ = intelligibility_enhancer_enabled;
     beamformer_enabled_ = beamformer_enabled;
     adaptive_gain_controller_enabled_ = adaptive_gain_controller_enabled;
     level_controller_enabled_ = level_controller_enabled;
     level_estimator_enabled_ = level_estimator_enabled;
     voice_activity_detector_enabled_ = voice_activity_detector_enabled;
     transient_suppressor_enabled_ = transient_suppressor_enabled;
   }
 
@@ skipping 54 matching lines @@
 #endif
 };
 
 struct AudioProcessingImpl::ApmPrivateSubmodules {
   explicit ApmPrivateSubmodules(NonlinearBeamformer* beamformer)
       : beamformer(beamformer) {}
   // Accessed internally from capture or during initialization
   std::unique_ptr<NonlinearBeamformer> beamformer;
   std::unique_ptr<AgcManagerDirect> agc_manager;
   std::unique_ptr<LevelController> level_controller;
+  std::unique_ptr<ResidualEchoDetector> residual_echo_detector;
 };
 
 AudioProcessing* AudioProcessing::Create() {
   webrtc::Config config;
   return Create(config, nullptr);
 }
 
 AudioProcessing* AudioProcessing::Create(const webrtc::Config& config) {
   return Create(config, nullptr);
 }
@@ skipping 45 matching lines @@
         new HighPassFilterImpl(&crit_capture_));
     public_submodules_->level_estimator.reset(
         new LevelEstimatorImpl(&crit_capture_));
     public_submodules_->noise_suppression.reset(
         new NoiseSuppressionImpl(&crit_capture_));
     public_submodules_->voice_detection.reset(
         new VoiceDetectionImpl(&crit_capture_));
     public_submodules_->gain_control_for_experimental_agc.reset(
         new GainControlForExperimentalAgc(
             public_submodules_->gain_control.get(), &crit_capture_));
+    private_submodules_->residual_echo_detector.reset(
+        new ResidualEchoDetector());
 
     // TODO(peah): Move this creation to happen only when the level controller
     // is enabled.
     private_submodules_->level_controller.reset(new LevelController());
   }
 
   SetExtraOptions(config);
 }
 
 AudioProcessingImpl::~AudioProcessingImpl() {
@@ skipping 145 matching lines @@
 #if WEBRTC_INTELLIGIBILITY_ENHANCER
   InitializeIntelligibility();
 #endif
   public_submodules_->high_pass_filter->Initialize(num_proc_channels(),
                                                    proc_sample_rate_hz());
   public_submodules_->noise_suppression->Initialize(num_proc_channels(),
                                                     proc_sample_rate_hz());
   public_submodules_->voice_detection->Initialize(proc_split_sample_rate_hz());
   public_submodules_->level_estimator->Initialize();
   InitializeLevelController();
+  InitializeResidualEchoDetector();
 
 #ifdef WEBRTC_AUDIOPROC_DEBUG_DUMP
   if (debug_dump_.debug_file->is_open()) {
     int err = WriteInitMessage();
     if (err != kNoError) {
       return err;
     }
   }
 #endif
 
@@ skipping 319 matching lines @@
     // Insert the samples into the queue.
     if (!agc_render_signal_queue_->Insert(&agc_render_queue_buffer_)) {
       // The data queue is full and needs to be emptied.
       EmptyQueuedRenderAudio();
 
       // Retry the insert (should always work).
       bool result = agc_render_signal_queue_->Insert(&agc_render_queue_buffer_);
       RTC_DCHECK(result);
     }
   }
+
+  ResidualEchoDetector::PackRenderAudioBuffer(audio, &red_render_queue_buffer_);
+
+  // Insert the samples into the queue.
+  if (!red_render_signal_queue_->Insert(&red_render_queue_buffer_)) {
+    // The data queue is full and needs to be emptied.
+    EmptyQueuedRenderAudio();
+
+    // Retry the insert (should always work).
+    bool result = red_render_signal_queue_->Insert(&red_render_queue_buffer_);
+    RTC_DCHECK(result);
+  }
 }
 
 void AudioProcessingImpl::AllocateRenderQueue() {
   const size_t new_aec_render_queue_element_max_size =
       std::max(static_cast<size_t>(1),
                kMaxAllowedValuesOfSamplesPerFrame *
                    EchoCancellationImpl::NumCancellersRequired(
                        num_output_channels(), num_reverse_channels()));
 
   const size_t new_aecm_render_queue_element_max_size =
       std::max(static_cast<size_t>(1),
                kMaxAllowedValuesOfSamplesPerFrame *
                    EchoControlMobileImpl::NumCancellersRequired(
                        num_output_channels(), num_reverse_channels()));
 
   const size_t new_agc_render_queue_element_max_size =
       std::max(static_cast<size_t>(1), kMaxAllowedValuesOfSamplesPerFrame);
 
+  const size_t new_red_render_queue_element_max_size =
+      std::max(static_cast<size_t>(1), kMaxAllowedValuesOfSamplesPerFrame);
+
   // Reallocate the queues if the queue item sizes are too small to fit the
   // data to put in the queues.
   if (aec_render_queue_element_max_size_ <
       new_aec_render_queue_element_max_size) {
     aec_render_queue_element_max_size_ = new_aec_render_queue_element_max_size;
 
     std::vector<float> template_queue_element(
         aec_render_queue_element_max_size_);
 
     aec_render_signal_queue_.reset(
@@ skipping 39 matching lines @@
         new SwapQueue<std::vector<int16_t>, RenderQueueItemVerifier<int16_t>>(
             kMaxNumFramesToBuffer, template_queue_element,
             RenderQueueItemVerifier<int16_t>(
                 agc_render_queue_element_max_size_)));
 
     agc_render_queue_buffer_.resize(agc_render_queue_element_max_size_);
     agc_capture_queue_buffer_.resize(agc_render_queue_element_max_size_);
   } else {
     agc_render_signal_queue_->Clear();
   }
+
+  if (red_render_queue_element_max_size_ <
+      new_red_render_queue_element_max_size) {
+    red_render_queue_element_max_size_ = new_red_render_queue_element_max_size;
+
+    std::vector<float> template_queue_element(
+        red_render_queue_element_max_size_);
+
+    red_render_signal_queue_.reset(
+        new SwapQueue<std::vector<float>, RenderQueueItemVerifier<float>>(
+            kMaxNumFramesToBuffer, template_queue_element,
+            RenderQueueItemVerifier<float>(
+                red_render_queue_element_max_size_)));
+
+    red_render_queue_buffer_.resize(red_render_queue_element_max_size_);
+    red_capture_queue_buffer_.resize(red_render_queue_element_max_size_);
+  } else {
+    red_render_signal_queue_->Clear();
+  }
 }
 
 void AudioProcessingImpl::EmptyQueuedRenderAudio() {
   rtc::CritScope cs_capture(&crit_capture_);
   while (aec_render_signal_queue_->Remove(&aec_capture_queue_buffer_)) {
     public_submodules_->echo_cancellation->ProcessRenderAudio(
         aec_capture_queue_buffer_);
   }
 
   while (aecm_render_signal_queue_->Remove(&aecm_capture_queue_buffer_)) {
     public_submodules_->echo_control_mobile->ProcessRenderAudio(
         aecm_capture_queue_buffer_);
   }
 
   while (agc_render_signal_queue_->Remove(&agc_capture_queue_buffer_)) {
     public_submodules_->gain_control->ProcessRenderAudio(
         agc_capture_queue_buffer_);
   }
+
+  while (red_render_signal_queue_->Remove(&red_capture_queue_buffer_)) {
+    private_submodules_->residual_echo_detector->AnalyzeRenderAudio(
+        red_capture_queue_buffer_);
+  }
 }
 
 int AudioProcessingImpl::ProcessStream(AudioFrame* frame) {
   TRACE_EVENT0("webrtc", "AudioProcessing::ProcessStream_AudioFrame");
   {
     // Acquire the capture lock in order to safely call the function
     // that retrieves the render side data. This function accesses apm
     // getters that need the capture lock held when being called.
     // The lock needs to be released as
     // public_submodules_->echo_control_mobile->is_enabled() aquires this lock
@@ skipping 155 matching lines @@
   // Ensure that the stream delay was set before the call to the
   // AECM ProcessCaptureAudio function.
   if (public_submodules_->echo_control_mobile->is_enabled() &&
       !was_stream_delay_set()) {
     return AudioProcessing::kStreamParameterNotSetError;
   }
 
   RETURN_ON_ERR(public_submodules_->echo_control_mobile->ProcessCaptureAudio(
       capture_buffer, stream_delay_ms()));
 
+  if (config_.residual_echo_detector.enabled) {
+    private_submodules_->residual_echo_detector->AnalyzeCaptureAudio(
+        rtc::ArrayView<const float>(
+            capture_buffer->split_bands_const_f(0)[kBand0To8kHz],
+            capture_buffer->num_frames_per_band()));
+  }
+
   if (capture_nonlocked_.beamformer_enabled) {
     private_submodules_->beamformer->PostFilter(capture_buffer->split_data_f());
   }
 
   public_submodules_->voice_detection->ProcessCaptureAudio(capture_buffer);
 
   if (constants_.use_experimental_agc &&
       public_submodules_->gain_control->is_enabled() &&
       (!capture_nonlocked_.beamformer_enabled ||
        private_submodules_->beamformer->is_target_present())) {
@@ skipping 355 matching lines @@
 
 VoiceDetection* AudioProcessingImpl::voice_detection() const {
   return public_submodules_->voice_detection.get();
 }
 
 bool AudioProcessingImpl::UpdateActiveSubmoduleStates() {
   return submodule_states_.Update(
       public_submodules_->high_pass_filter->is_enabled(),
       public_submodules_->echo_cancellation->is_enabled(),
       public_submodules_->echo_control_mobile->is_enabled(),
+      config_.residual_echo_detector.enabled,
       public_submodules_->noise_suppression->is_enabled(),
       capture_nonlocked_.intelligibility_enabled,
       capture_nonlocked_.beamformer_enabled,
       public_submodules_->gain_control->is_enabled(),
       capture_nonlocked_.level_controller_enabled,
       public_submodules_->voice_detection->is_enabled(),
       public_submodules_->level_estimator->is_enabled(),
       capture_.transient_suppressor_enabled);
 }
 
@@ skipping 29 matching lines @@
                                     render_.render_audio->num_bands(),
                                     NoiseSuppressionImpl::num_noise_bins()));
   }
 #endif
 }
 
 void AudioProcessingImpl::InitializeLevelController() {
   private_submodules_->level_controller->Initialize(proc_sample_rate_hz());
 }
 
+void AudioProcessingImpl::InitializeResidualEchoDetector() {
+  private_submodules_->residual_echo_detector->Initialize();
+}
+
 void AudioProcessingImpl::MaybeUpdateHistograms() {
   static const int kMinDiffDelayMs = 60;
 
   if (echo_cancellation()->is_enabled()) {
     // Activate delay_jumps_ counters if we know echo_cancellation is runnning.
     // If a stream has echo we know that the echo_cancellation is in process.
     if (capture_.stream_delay_jumps == -1 &&
         echo_cancellation()->stream_has_echo()) {
       capture_.stream_delay_jumps = 0;
     }
@@ skipping 218 matching lines @@
       capture_processing_format(kSampleRate16kHz),
       split_rate(kSampleRate16kHz) {}
 
 AudioProcessingImpl::ApmCaptureState::~ApmCaptureState() = default;
 
 AudioProcessingImpl::ApmRenderState::ApmRenderState() = default;
 
 AudioProcessingImpl::ApmRenderState::~ApmRenderState() = default;
 
 }  // namespace webrtc