move the APM to chrome.

move the APM to chrome.

BUG=

[perkj_chrome, 2013-10-24 12:31:30]

Nice hacking.  A few initial comments.

Also what we talked about- the feedback method should probably be an
implementation of an interface so that the provider of the feedback don't need
to know about the source.

https://codereview.chromium.org/37793005/diff/1/content/renderer/media/webrtc...
File content/renderer/media/webrtc_audio_processing_wrapper.cc (right):

https://codereview.chromium.org/37793005/diff/1/content/renderer/media/webrtc...
content/renderer/media/webrtc_audio_processing_wrapper.cc:207: const
MediaConstraintsInterface* constraints) {
You might want to consider mandatory and optional constraints. Ie- If there are
mandatory constraints in |constraints| that we can not fullfill- Configure
should fail.

Just add a todo for now.

https://codereview.chromium.org/37793005/diff/1/content/renderer/media/webrtc...
content/renderer/media/webrtc_audio_processing_wrapper.cc:273:
capture_converter_->Push(audio_source);
What if all features of APM is disabled- does this still work?

[Henrik Grunell, 2013-10-31 11:56:11]

I'm assuming that
- the description of the issue will be updated once the design is more stable,
- a unit test will be added for the audio processor.

Some random comments below.

https://codereview.chromium.org/37793005/diff/100001/content/public/common/co...
File content/public/common/content_switches.cc (right):

https://codereview.chromium.org/37793005/diff/100001/content/public/common/co...
content/public/common/content_switches.cc:917: const char
kEnableWebRtcAudioProcessor[]    = "enable-audio-processor";
I think the name should reflect that the switch changes where processing takes
place. I can't think of a very good name, maybe something like
"switch-audio-processing-from-webrtc-to-chrome"? Or
"use-audio-processing-per-track"?

https://codereview.chromium.org/37793005/diff/100001/content/renderer/media/w...
File content/renderer/media/webrtc_audio_device_impl.h (right):

https://codereview.chromium.org/37793005/diff/100001/content/renderer/media/w...
content/renderer/media/webrtc_audio_device_impl.h:213: virtual int
CaptureData(const std::vector<int>& channels,
Why remove the description of some input parameters and the return value?

https://codereview.chromium.org/37793005/diff/100001/content/renderer/media/w...
content/renderer/media/webrtc_audio_device_impl.h:230: class
WebRtcAudioRenderDataObserver {
Add comment. See also next comment.

https://codereview.chromium.org/37793005/diff/100001/content/renderer/media/w...
content/renderer/media/webrtc_audio_device_impl.h:318: // Add/Remove the
observer to get the mixed render data for AEC.
I think the description "to get the mixed render data for AEC" should be moved
to a comment for WebRtcAudioRenderDataObserver above. Also change the text so
that it's clear that using the data for AEC is one example. There may be other
uses in the future.

https://codereview.chromium.org/37793005/diff/100001/content/renderer/media/w...
File content/renderer/media/webrtc_audio_processor.cc (right):

https://codereview.chromium.org/37793005/diff/100001/content/renderer/media/w...
content/renderer/media/webrtc_audio_processor.cc:99: static const char
kAecDumpFilename[] = "/tmp/audio.aecdump";
This should be different for different platforms. See webrtcvoiceengine.cc in
libjingle ~line 138. Also move the comment and todo (for crbug.com/264223) here.

https://codereview.chromium.org/37793005/diff/100001/content/renderer/media/w...
File content/renderer/media/webrtc_audio_processor.h (right):

https://codereview.chromium.org/37793005/diff/100001/content/renderer/media/w...
content/renderer/media/webrtc_audio_processor.h:37: void Push(media::AudioBus*
audio_source);
Add comment.