Adding audio RepetitionDetector in AudioProcessingModule.

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 14 matching lines @@
 #include "webrtc/modules/audio_processing/audio_buffer.h"
 #include "webrtc/modules/audio_processing/beamformer/nonlinear_beamformer.h"
 #include "webrtc/modules/audio_processing/common.h"
 #include "webrtc/modules/audio_processing/echo_cancellation_impl.h"
 #include "webrtc/modules/audio_processing/echo_control_mobile_impl.h"
 #include "webrtc/modules/audio_processing/gain_control_impl.h"
 #include "webrtc/modules/audio_processing/high_pass_filter_impl.h"
 #include "webrtc/modules/audio_processing/level_estimator_impl.h"
 #include "webrtc/modules/audio_processing/noise_suppression_impl.h"
 #include "webrtc/modules/audio_processing/processing_component.h"
+#include "webrtc/modules/audio_processing/repetition_detector.h"
 #include "webrtc/modules/audio_processing/transient/transient_suppressor.h"
 #include "webrtc/modules/audio_processing/voice_detection_impl.h"
 #include "webrtc/modules/interface/module_common_types.h"
 #include "webrtc/system_wrappers/interface/critical_section_wrapper.h"
 #include "webrtc/system_wrappers/interface/file_wrapper.h"
 #include "webrtc/system_wrappers/interface/logging.h"
 #include "webrtc/system_wrappers/interface/metrics.h"
 
 #ifdef WEBRTC_AUDIOPROC_DEBUG_DUMP
 // Files generated at build-time by the protobuf compiler.
@@ skipping 158 matching lines @@
       use_new_agc_(config.Get<ExperimentalAgc>().enabled),
 #endif
       agc_startup_min_volume_(config.Get<ExperimentalAgc>().startup_min_volume),
 #if defined(WEBRTC_ANDROID) || defined(WEBRTC_IOS)
       transient_suppressor_enabled_(false),
 #else
       transient_suppressor_enabled_(config.Get<ExperimentalNs>().enabled),
 #endif
       beamformer_enabled_(config.Get<Beamforming>().enabled),
       beamformer_(beamformer),
-      array_geometry_(config.Get<Beamforming>().array_geometry) {
+      array_geometry_(config.Get<Beamforming>().array_geometry),
+      repetition_detector_(new RepetitionDetector()) {

[Andrew MacDonald, 2015/08/11 18:53:05]

We usually lazily initialize components, so you would create this only if
requested.

[minyue-webrtc, 2015/08/12 11:32:13]

Acknowledged.

   echo_cancellation_ = new EchoCancellationImpl(this, crit_);
   component_list_.push_back(echo_cancellation_);
 
   echo_control_mobile_ = new EchoControlMobileImpl(this, crit_);
   component_list_.push_back(echo_control_mobile_);
 
   gain_control_ = new GainControlImpl(this, crit_);
   component_list_.push_back(gain_control_);
 
   high_pass_filter_ = new HighPassFilterImpl(this, crit_);
@@ skipping 338 matching lines @@
       frame->sample_rate_hz_ != kSampleRate32kHz &&
       frame->sample_rate_hz_ != kSampleRate48kHz) {
     return kBadSampleRateError;
   }
   if (echo_control_mobile_->is_enabled() &&
       frame->sample_rate_hz_ > kSampleRate16kHz) {
     LOG(LS_ERROR) << "AECM only supports 16 or 8 kHz sample rates";
     return kUnsupportedComponentError;
   }
 
+  repetition_detector_->Detect(frame->data_, frame->samples_per_channel_,

[Andrew MacDonald, 2015/08/11 20:50:42]

All processing/analysis should be done in ProcessStreamLocked.

[minyue-webrtc, 2015/08/12 11:32:13]

We'd better access raw data (frame), ProcessStreamLocked() does not have access
to it. What would you suggest?

[Andrew MacDonald, 2015/08/12 20:25:32]

See how the other components access the data (through AudioBuffer). You're right
that resampling could occur before you get access to the audio, but that could
have happened outside APM already anyway. I'd attempt to make your algorithm
robust to minor variations.

[peah-webrtc, 2015/08/12 21:05:16]

That makes sense. I today looked into a recording where the matching was not
100% identical sample wise but where the difference was tiny and it was clearly
repeated audio.

+                               frame->num_channels_);
+
   // TODO(ajm): The input and output rates and channels are currently
   // constrained to be identical in the int16 interface.
   ProcessingConfig processing_config = api_format_;
   processing_config.input_stream().set_sample_rate_hz(frame->sample_rate_hz_);
   processing_config.input_stream().set_num_channels(frame->num_channels_);
   processing_config.output_stream().set_sample_rate_hz(frame->sample_rate_hz_);
   processing_config.output_stream().set_num_channels(frame->num_channels_);
 
   RETURN_ON_ERR(MaybeInitializeLocked(processing_config));
   if (frame->samples_per_channel_ != api_format_.input_stream().num_frames()) {
@@ skipping 34 matching lines @@
     msg->set_delay(stream_delay_ms_);
     msg->set_drift(echo_cancellation_->stream_drift_samples());
     msg->set_level(gain_control()->stream_analog_level());
     msg->set_keypress(key_pressed_);
   }
 #endif
 
   MaybeUpdateHistograms();
 
   AudioBuffer* ca = capture_audio_.get();  // For brevity.
+

[hlundin-webrtc, 2015/08/12 14:12:04]

Why new line?

   if (use_new_agc_ && gain_control_->is_enabled()) {
     agc_manager_->AnalyzePreProcess(ca->channels()[0], ca->num_channels(),
                                     fwd_proc_format_.num_frames());
   }
 
   bool data_processed = is_data_processed();
   if (analysis_needed(data_processed)) {
     ca->SplitIntoFrequencyBands();
   }
 
@@ skipping 504 matching lines @@
   int err = WriteMessageToDebugFile();
   if (err != kNoError) {
     return err;
   }
 
   return kNoError;
 }
 #endif  // WEBRTC_AUDIOPROC_DEBUG_DUMP
 
 }  // namespace webrtc