AecDump interface

webrtc/modules/audio_processing/audio_processing_impl.cc

Index: webrtc/modules/audio_processing/audio_processing_impl.cc
diff --git a/webrtc/modules/audio_processing/audio_processing_impl.cc b/webrtc/modules/audio_processing/audio_processing_impl.cc
index 816210f34f8738eb7a042adee69f87f5ed1a99d8..85a3e6003b4c6256e4f1df916267fffec6e075a1 100644
--- a/webrtc/modules/audio_processing/audio_processing_impl.cc
+++ b/webrtc/modules/audio_processing/audio_processing_impl.cc
@@ -129,6 +129,9 @@ static const size_t kMaxAllowedValuesOfSamplesPerFrame = 160;
 // reverse and forward call numbers.
 static const size_t kMaxNumFramesToBuffer = 100;
 
+// Maximum number of audio channels in the input and output streams.
+constexpr size_t kMaxNumChannels = 4;

[peah-webrtc, 2017/04/21 14:05:15]

I think you commented that this would be changed.

[aleloi, 2017/04/24 11:10:09]

The name or the number? I've changed the name now (somehow the change didn't end
up in this P-set). I think we said that 4 should be enough for now.

+
 class HighPassFilterImpl : public HighPassFilter {
  public:
   explicit HighPassFilterImpl(AudioProcessingImpl* apm) : apm_(apm) {}
@@ -152,6 +155,37 @@ class HighPassFilterImpl : public HighPassFilter {
   RTC_DISALLOW_IMPLICIT_CONSTRUCTORS(HighPassFilterImpl);
 };
 
+webrtc::InternalAPMStreamsConfig ToStreamsConfig(
+    const ProcessingConfig& api_format) {
+  webrtc::InternalAPMStreamsConfig result;
+  result.input_sample_rate = api_format.input_stream().sample_rate_hz();
+  result.input_num_channels = api_format.input_stream().num_channels();
+  result.output_num_channels = api_format.output_stream().num_channels();
+  result.render_input_num_channels =
+      api_format.reverse_input_stream().num_channels();
+  result.render_input_sample_rate =
+      api_format.reverse_input_stream().sample_rate_hz();
+  result.output_sample_rate = api_format.output_stream().sample_rate_hz();
+  result.render_output_sample_rate =
+      api_format.reverse_output_stream().sample_rate_hz();
+  result.render_output_num_channels =
+      api_format.reverse_output_stream().num_channels();
+  return result;
+}
+
+rtc::ArrayView<rtc::ArrayView<const float>> stream_view(

[peah-webrtc, 2017/04/21 14:05:15]

Does this name follow the style guideline? What about CreateStreamView?

[aleloi, 2017/04/24 11:10:09]

I thought the style allowed names_with_underscores for cheap functions.
Re-reading it showed I was wrong:

  Regular functions have mixed case; accessors and mutators may be named like
variables.


In any case, CreateStreamView is more descriptive.

+    const float* const* stream,
+    size_t channel_size,
+    size_t num_channels) {
+  RTC_DCHECK_LE(num_channels, kMaxNumChannels);
+
+  std::array<rtc::ArrayView<const float>, kMaxNumChannels> array_stream_view;
+  for (size_t i = 0; i < std::min(num_channels, kMaxNumChannels); ++i) {
+    array_stream_view[i] = rtc::ArrayView<const float>(stream[i], channel_size);
+  }
+  return rtc::ArrayView<rtc::ArrayView<const float>>(&array_stream_view[0],
+                                                     num_channels);
+}
 }  // namespace
 
 // Throughout webrtc, it's assumed that success is represented by zero.
@@ -526,7 +560,9 @@ int AudioProcessingImpl::InitializeLocked() {
     }
   }
 #endif
-
+  if (aec_dump_) {
+    aec_dump_->WriteInitMessage(ToStreamsConfig(formats_.api_format));
+  }
   return kNoError;
 }
 
@@ -824,6 +860,20 @@ int AudioProcessingImpl::ProcessStream(const float* const* src,
   }
 #endif
 
+  std::unique_ptr<AecDump::CaptureStreamInfo> stream_info;
+  if (aec_dump_) {
+    aec_dump_->WriteConfig(CollectApmConfig(), false);

[peah-webrtc, 2017/04/21 14:05:15]

Same thing here, would it be possible to move the whole recording functionality
into a method?

[aleloi, 2017/04/24 11:10:09]

WDYT about the latest change? I don't think it's perfect, because the
RecordUnprocessedCaptureStream() method does lots of things at once, but at
least ProcessStream looks simpler now.

+
+    stream_info = aec_dump_->GetCaptureStreamInfo();
+    RTC_DCHECK(stream_info);
+    const size_t channel_size =
+        sizeof(float) * formats_.api_format.input_stream().num_frames();
+    const size_t num_channels =
+        formats_.api_format.input_stream().num_channels();
+    stream_info->AddInput(stream_view(src, channel_size, num_channels));
+    PopulateStreamInfoWithConfig(stream_info.get());
+  }
+
   capture_.capture_audio->CopyFrom(src, formats_.api_format.input_stream());
   RETURN_ON_ERR(ProcessCaptureStreamLocked());
   capture_.capture_audio->CopyTo(formats_.api_format.output_stream(), dest);
@@ -841,6 +891,14 @@ int AudioProcessingImpl::ProcessStream(const float* const* src,
                                           &crit_debug_, &debug_dump_.capture));
   }
 #endif
+  if (aec_dump_) {
+    const size_t channel_size =
+        sizeof(float) * formats_.api_format.output_stream().num_frames();

[peah-webrtc, 2017/04/21 14:05:15]

Same thing here, would it be possible to move the whole recording functionality
into a method?

[aleloi, 2017/04/24 11:10:09]

Acknowledged.

+    const size_t num_channels =
+        formats_.api_format.output_stream().num_channels();
+    stream_info->AddOutput(stream_view(dest, channel_size, num_channels));
+    aec_dump_->WriteCaptureStreamMessage(std::move(stream_info));
+  }
 
   return kNoError;
 }
@@ -1078,6 +1136,15 @@ int AudioProcessingImpl::ProcessStream(AudioFrame* frame) {
     return kBadDataLengthError;
   }
 
+  std::unique_ptr<AecDump::CaptureStreamInfo> stream_info;
+  if (aec_dump_) {
+    stream_info = aec_dump_->GetCaptureStreamInfo();

[peah-webrtc, 2017/04/21 14:05:15]

Same thing here, would it be possible to move the whole recording functionality
into a method?

[aleloi, 2017/04/24 11:10:09]

Acknowledged.

+    RTC_DCHECK(stream_info);
+    stream_info->AddInput(*frame);
+    PopulateStreamInfoWithConfig(stream_info.get());
+    aec_dump_->WriteConfig(CollectApmConfig(), false);
+  }
+
 #ifdef WEBRTC_AUDIOPROC_DEBUG_DUMP
   if (debug_dump_.debug_file->is_open()) {
     RETURN_ON_ERR(WriteConfigMessage(false));
@@ -1095,6 +1162,10 @@ int AudioProcessingImpl::ProcessStream(AudioFrame* frame) {
   capture_.capture_audio->InterleaveTo(
       frame, submodule_states_.CaptureMultiBandProcessingActive());
 
+  if (aec_dump_) {
+    stream_info->AddOutput(*frame);
+    aec_dump_->WriteCaptureStreamMessage(std::move(stream_info));
+  }
 #ifdef WEBRTC_AUDIOPROC_DEBUG_DUMP
   if (debug_dump_.debug_file->is_open()) {
     audioproc::Stream* msg = debug_dump_.capture.event_msg->mutable_stream();
@@ -1376,7 +1447,14 @@ int AudioProcessingImpl::AnalyzeReverseStreamLocked(
                                           &crit_debug_, &debug_dump_.render));
   }
 #endif
-
+  if (aec_dump_) {
+    const size_t channel_size =
+        sizeof(float) * formats_.api_format.reverse_input_stream().num_frames();
+    const size_t num_channels =

[peah-webrtc, 2017/04/21 14:05:15]

Same thing here, would it be possible to move the whole recording functionality
into a method?

[aleloi, 2017/04/24 11:10:09]

Acknowledged.

+        formats_.api_format.reverse_input_stream().num_channels();
+    aec_dump_->WriteRenderStreamMessage(
+        stream_view(src, channel_size, num_channels));
+  }
   render_.render_audio->CopyFrom(src,
                                  formats_.api_format.reverse_input_stream());
   return ProcessRenderStreamLocked();
@@ -1429,6 +1507,10 @@ int AudioProcessingImpl::ProcessReverseStream(AudioFrame* frame) {
                                           &crit_debug_, &debug_dump_.render));
   }
 #endif
+  if (aec_dump_) {
+    aec_dump_->WriteRenderStreamMessage(*frame);
+  }
+
   render_.render_audio->DeinterleaveFrom(frame);
   RETURN_ON_ERR(ProcessRenderStreamLocked());
   render_.render_audio->InterleaveTo(
@@ -1512,6 +1594,17 @@ int AudioProcessingImpl::delay_offset_ms() const {
   return capture_.delay_offset_ms;
 }
 
+void AudioProcessingImpl::StartDebugRecording(
+    std::unique_ptr<AecDump> aec_dump) {
+  rtc::CritScope cs_render(&crit_render_);
+  rtc::CritScope cs_capture(&crit_capture_);
+  RTC_DCHECK(aec_dump);
+  aec_dump_ = std::move(aec_dump);
+
+  aec_dump_->WriteConfig(CollectApmConfig(), true);
+  aec_dump_->WriteInitMessage(ToStreamsConfig(formats_.api_format));
+}
+
 int AudioProcessingImpl::StartDebugRecording(
     const char filename[AudioProcessing::kMaxFilenameSize],
     int64_t max_log_size_bytes) {
@@ -1586,6 +1679,7 @@ int AudioProcessingImpl::StopDebugRecording() {
   // Run in a single-threaded manner.
   rtc::CritScope cs_render(&crit_render_);
   rtc::CritScope cs_capture(&crit_capture_);
+  aec_dump_.reset();
 
 #ifdef WEBRTC_AUDIOPROC_DEBUG_DUMP
   // We just return if recording hasn't started.
@@ -1837,6 +1931,71 @@ void AudioProcessingImpl::UpdateHistogramsOnCallEnd() {
   capture_.last_aec_system_delay_ms = 0;
 }
 
+InternalAPMConfig AudioProcessingImpl::CollectApmConfig() {
+  std::string experiments_description =
+      public_submodules_->echo_cancellation->GetExperimentsDescription();
+  // TODO(peah): Add semicolon-separated concatenations of experiment
+  // descriptions for other submodules.
+  if (capture_nonlocked_.level_controller_enabled) {
+    experiments_description += "LevelController;";
+  }
+  if (constants_.agc_clipped_level_min != kClippedLevelMin) {
+    experiments_description += "AgcClippingLevelExperiment;";
+  }
+  if (capture_nonlocked_.echo_canceller3_enabled) {
+    experiments_description += "EchoCanceller3;";
+  }
+
+  InternalAPMConfig apm_config;
+
+  apm_config.aec_enabled = public_submodules_->echo_cancellation->is_enabled();
+  apm_config.aec_delay_agnostic_enabled =
+      public_submodules_->echo_cancellation->is_delay_agnostic_enabled();
+  apm_config.aec_drift_compensation_enabled =
+      public_submodules_->echo_cancellation->is_drift_compensation_enabled();
+  apm_config.aec_extended_filter_enabled =
+      public_submodules_->echo_cancellation->is_extended_filter_enabled();
+  apm_config.aec_suppression_level = static_cast<int>(
+      public_submodules_->echo_cancellation->suppression_level());
+
+  apm_config.aecm_enabled =
+      public_submodules_->echo_control_mobile->is_enabled();
+  apm_config.aecm_comfort_noise_enabled =
+      public_submodules_->echo_control_mobile->is_comfort_noise_enabled();
+  apm_config.aecm_routing_mode =
+      static_cast<int>(public_submodules_->echo_control_mobile->routing_mode());
+
+  apm_config.agc_enabled = public_submodules_->gain_control->is_enabled();
+  apm_config.agc_mode =
+      static_cast<int>(public_submodules_->gain_control->mode());
+  apm_config.agc_limiter_enabled =
+      public_submodules_->gain_control->is_limiter_enabled();
+  apm_config.noise_robust_agc_enabled = constants_.use_experimental_agc;
+
+  apm_config.hpf_enabled = config_.high_pass_filter.enabled;
+
+  apm_config.ns_enabled = public_submodules_->noise_suppression->is_enabled();
+  apm_config.ns_level =
+      static_cast<int>(public_submodules_->noise_suppression->level());
+
+  apm_config.transient_suppression_enabled =
+      capture_.transient_suppressor_enabled;
+  apm_config.intelligibility_enhancer_enabled =
+      capture_nonlocked_.intelligibility_enabled;
+  apm_config.experiments_description = experiments_description;
+  return apm_config;
+}
+
+void AudioProcessingImpl::PopulateStreamInfoWithConfig(
+    AecDump::CaptureStreamInfo* stream_info) const {
+  RTC_DCHECK(stream_info);
+  stream_info->set_delay(capture_nonlocked_.stream_delay_ms);
+  stream_info->set_drift(
+      public_submodules_->echo_cancellation->stream_drift_samples());
+  stream_info->set_level(gain_control()->stream_analog_level());
+  stream_info->set_keypress(capture_.key_pressed);
+}
+
 #ifdef WEBRTC_AUDIOPROC_DEBUG_DUMP
 int AudioProcessingImpl::WriteMessageToDebugFile(
     FileWrapper* debug_file,