MediaStream audio sourcing: Bypass audio processing for non-WebRTC cases.

MediaStream audio sourcing: Bypass audio processing for non-WebRTC cases.

This change adds a second MediaStreamTrack implementation for directly
delivering audio from the source to the sinks without any processing or
FIFO in-between.  This is to correct severe audio distortions present
in the following use cases: 1) local "loopback" audio associated with
screen capture; 2) remotely-received audio via the Cast Streaming
Receiver; 3) any other audio source, local or remote, where there are no
constraints indicating audio processing need take place.

BUG=577881, 577874

[miu, 2016-01-28 01:04:02]

Patchset #1 (id:1) has been deleted

[miu, 2016-01-28 20:59:57]

Description was changed from

==========
MediaStream audio sourcing: Bypass audio processing for non-WebRTC cases.

<<TODO: ENTER DESCRIPTION HERE>>

BUG=577881, 577874
==========

to

==========
MediaStream audio sourcing: Bypass audio processing for non-WebRTC cases.

This change adds a second MediaStreamTrack implementation for directly
delivering audio from the source to the sinks without any processing or
FIFO in-between.  This is to correct severe audio distortions present
in the following use cases: 1) local "loopback" audio associated with
screen capture; 2) remotely-received audio via the Cast Streaming
Receiver; 3) any other audio source, local or remote, where there are no
constraints indicating audio processing need take place.

BUG=577881, 577874
==========

[miu, 2016-01-28 21:08:29]

miu@chromium.org changed reviewers:
+ hubbe@chromium.org

[miu, 2016-01-28 21:08:30]

hubbe: PTAL.  Whereas the other CL is about audio output, this one is about
audio input.

[miu, 2016-01-28 21:16:30]

miu@chromium.org changed reviewers:
+ grunell@chromium.org, tommi@chromium.org

[miu, 2016-01-28 21:16:30]

Adding tommi@ and grunell@: PTAL (CL 2 of 2).  These two CLs have similar
titles, but the first one is about rendering audio unmodified while this one is
about delivering audio unmodified from source to sinks.

[hubbe, 2016-01-28 21:48:13]

https://codereview.chromium.org/1647773002/diff/20001/content/public/common/m...
File content/public/common/media_stream_request.h (right):

https://codereview.chromium.org/1647773002/diff/20001/content/public/common/m...
content/public/common/media_stream_request.h:90: CONTENT_EXPORT bool
IsContentCaptureMediaType(MediaStreamType type);
Can I haz comment?

https://codereview.chromium.org/1647773002/diff/20001/content/renderer/media/...
File content/renderer/media/media_stream_audio_source.h (right):

https://codereview.chromium.org/1647773002/diff/20001/content/renderer/media/...
content/renderer/media/media_stream_audio_source.h:44: class CONTENT_EXPORT
MediaStreamAudioSource
Would it make sense to write this as an abstract class with three different
concrete implementations?

https://codereview.chromium.org/1647773002/diff/20001/content/renderer/media/...
content/renderer/media/media_stream_audio_source.h:151: NULL_SOURCE_NOT_STARTED,
  // This instance is a "null" audio source.
I'm not a big fan of using different states for different types of inputs.
Seems like a mix of purposes.

A single "NOT_STARTED", should be enough I think, and the type of input can
either be represented in a separate variable, or by using one subclass per input
type.

[Henrik Grunell, 2016-01-29 09:56:15]

grunell@chromium.org changed reviewers:
+ olka@chromium.org

[Henrik Grunell, 2016-01-29 09:56:27]

Adding olka@ as reviewer.

[miu, 2016-01-29 19:43:59]

https://codereview.chromium.org/1647773002/diff/20001/content/renderer/media/...
File content/renderer/media/media_stream_audio_source.h (right):

https://codereview.chromium.org/1647773002/diff/20001/content/renderer/media/...
content/renderer/media/media_stream_audio_source.h:44: class CONTENT_EXPORT
MediaStreamAudioSource
On 2016/01/28 21:48:13, hubbe wrote:
> Would it make sense to write this as an abstract class with three different
> concrete implementations?

Good point.  I'll work on this and ping when ready for review again.

[miu, 2016-02-03 03:53:59]

All: Please hold on further review.  I am still working on a new patch set with
a much cleaner refactoring.  Will ping when ready to continue.

[miu, 2016-02-18 23:48:02]

Patchset #3 (id:60001) has been deleted

[miu, 2016-02-18 23:48:10]

Patchset #2 (id:40001) has been deleted

[miu, 2016-02-18 23:50:09]

I'm closing this issue.  The latest patchset has a snapshot of a mostly-complete
refactoring, but please don't feel the need to review all this (in particular, I
can't seem to get git to realize the file renames very well).  I'll be breaking
this up into several changes that focus on specific issues so that all the
reshuffling makes sense.