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..81c30fba48141cd9200407eb39febe59f599c70c 100644
--- a/webrtc/voice_engine/output_mixer.cc
+++ b/webrtc/voice_engine/output_mixer.cc
@@ -586,6 +586,7 @@ void OutputMixer::APMAnalyzeReverseStream() {
     WEBRTC_TRACE(kTraceWarning, kTraceVoice, VoEId(_instanceId,-1),
                  "AudioProcessingModule::AnalyzeReverseStream() => error");
   }
+  RemixAndResample(frame, &audioproc_resampler_, &_audioFrame);

[Andrew MacDonald, 2015/07/24 23:50:40]

This is a big deal ;) First, you can't use the same resampler here, as it should
be operating on the same continuous stream.

I think the right way to do this is remove the pre-RemixAndResample and let APM
handle it. In the "modern" float interface that would be no problem.
Unfortunately, this AudioFrame interface doesn't handle resampling in
DeinterleaveFrom, as you've seen. So we'd need to add support for it there.

Another, and possibly superior, option is to remove support for IE through the
AudioFrame interface in this CL.

[ekm, 2015/07/29 00:37:19]

Done (1st option): added support in InterleaveTo in audio_buffer.

 }
 
 int