Integrate Intelligibility with APM

webrtc/voice_engine/output_mixer.cc

Index: webrtc/voice_engine/output_mixer.cc
diff --git a/webrtc/voice_engine/output_mixer.cc b/webrtc/voice_engine/output_mixer.cc
index 7cf98d390cb93afb068cd679563386ace2c9288d..62de841e4eba5ad34327efe2157135b919140169 100644
--- a/webrtc/voice_engine/output_mixer.cc
+++ b/webrtc/voice_engine/output_mixer.cc
@@ -575,14 +575,7 @@ OutputMixer::DoOperationsOnCombinedSignal(bool feed_data_to_apm)
 // ----------------------------------------------------------------------------
 
 void OutputMixer::APMAnalyzeReverseStream() {
-  // Convert from mixing to AudioProcessing sample rate, determined by the send
-  // side. Downmix to mono.
-  AudioFrame frame;
-  frame.num_channels_ = 1;
-  frame.sample_rate_hz_ = _audioProcessingModulePtr->input_sample_rate_hz();
-  RemixAndResample(_audioFrame, &audioproc_resampler_, &frame);
-
-  if (_audioProcessingModulePtr->AnalyzeReverseStream(&frame) == -1) {
+  if (_audioProcessingModulePtr->AnalyzeReverseStream(&_audioFrame) == -1) {

[Andrew MacDonald, 2015/07/29 03:52:28]

Can you run through a bunch of codecs in voe_cmd_test with AEC on/off
intelligibility on/off and make sure everything sounds OK? I'll do the same once
you land this.

[ekm, 2015/07/29 23:35:06]

Tried a bunch on loopback. AEC is effective for all with or without
intelligibility enabled, except for opus. For opus, I couldn't get AEC to be
effective in any case, even using AnalyzeReverseStream instead of
ProcessReverseStream. I'm guessing this is a separate issue.

[Andrew MacDonald, 2015/07/30 03:53:16]

Hmm, OK, I'll try it once this lands.

[ekm, 2015/07/30 06:15:18]

Acknowledged.

     WEBRTC_TRACE(kTraceWarning, kTraceVoice, VoEId(_instanceId,-1),
                  "AudioProcessingModule::AnalyzeReverseStream() => error");
   }