Allow intelligibility to compile in apm

Allow intelligibility to compile in apm

- Added files to gyp and BUILD
- Made minor fixes to get everything to compile
    and intelligibility_proc to run
- Added comments
- Auto-reformatting

Original cl is at: https://webrtc-codereview.appspot.com/57579004/

TBR=aluebs@webrtc.org

Committed: https://chromium.googlesource.com/external/webrtc/+/b7553dfdbb1ca7779eb0d80b5f509523c9b00086

BUG=

[ekm, 2015-06-17 01:57:43]

Committed patchset #1 (id:1) manually as
b7553dfdbb1ca7779eb0d80b5f509523c9b00086 (presubmit successful).

[aluebs-webrtc, 2015-06-17 03:25:56]

A revert of this CL (patchset #1 id:1) has been created in
https://codereview.webrtc.org/1187033005/ by aluebs@webrtc.org.

The reason for reverting is: Breaking the build bots:
http://build.chromium.org/p/client.webrtc/builders/Mac32%20Release%20%5Blarge...

Fails to compile with this error:

../../webrtc/modules/audio_processing/intelligibility/intelligibility_enhancer.cc:218:25:
error: no member named 'accumulate' in namespace 'std'
    power_target = std::accumulate(clear_variance_.variance(),.

[aluebs-google, 2015-06-17 03:29:03]

aluebs@google.com changed reviewers:
+ aluebs@google.com

[aluebs-google, 2015-06-17 03:29:04]

Else it would have lgtm'd...

[aluebs-webrtc, 2015-06-17 03:29:43]

Else it would have lgtm'd...

[ekm, 2015-06-17 17:12:45]

Patchset #2 (id:20001) has been deleted

[ekm, 2015-06-17 19:50:50]

Patchset #2 (id:40001) has been deleted

[ekm, 2015-06-17 23:00:40]

ekmeyerson@webrtc.org changed reviewers:
+ andrew@webrtc.org, turajs@google.com

[ekm, 2015-06-17 23:10:03]

Added <numeric> header to intelligibility_enhancer.cc which fixed all trys
except windows.

intelligibility_proc.cc had several things not compatible with windows, so I
followed Andrew's advice to use WavReader/Writer instead of 'esoteric AU formats
and memory mapping'
(https://webrtc-codereview.appspot.com/57579004/diff/1/webrtc/modules/audio_pr...).

[Andrew MacDonald, 2015-06-18 03:38:15]

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
File webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc
(right):

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:33:
const int16_t* in_ipcm;
I don't think any of these from here down to line 43 need to be globals, right?

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:85:
const std::string kTmpOutName = "tmp_out_file.wav";
Prefer not to use non-POD types as static globals. Either make this a const
char* or move it into void_main.

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:103:
std::string noise_file_name = noise_name;
You have three variables referring to each of the filenames. I think you can get
away with just using the FLAGS variant everywhere.

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:111:
ASSERT_EQ(stat(in_name, &in_stat), 0) << "Empty speech input.";
Not sure these are really needed. How often will you have any empty file?

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:123:
in_fpcm = new float[samples];
vector or at least scoped_ptr. Doesn't matter if we leak here of course, but
let's practice good habits ;)

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:127:
noise_fpcm = new float[samples];
vector

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:130:
noise_fpcm[i] = noise_fpcm[i % noise_samples];
This is a bit weird. You're restarting the noise file, which may not be a great
idea, depending on the content.

I think better to crop to the shorter of the two files.

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:148: for
(size_t i = 0; i < samples; i += fragment_size) {
I think it's more typical to read directly from the file in chunks, but this is
OK.

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:156:
WavWriter out_file(out_file_name, FLAGS_sample_rate, kNumChannels);
Indent inside scopes.

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:165: }
while (FLAGS_repeat);
This probably made sense with the piping directly to aplay. But now that you
write the whole thing to file then read, you're doing to have a gap between the
repeats. I think better to just remove this option.

Not sure this is worth it, but if you want the same behavior, you could open a
fifo:
http://linux.die.net/man/3/mkfifo

Write to it using WavWriter and read from it using aplay.

[Andrew MacDonald, 2015-06-18 03:39:22]

Oh, I just noticed, why is this CL not closed? Did you reopen it?

You should start a new CL for your attempt at relanding.

[ekm, 2015-06-18 18:50:27]

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
File webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc
(right):

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:33:
const int16_t* in_ipcm;
On 2015/06/18 03:38:15, andrew wrote:
> I don't think any of these from here down to line 43 need to be globals,
right?

Done.

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:85:
const std::string kTmpOutName = "tmp_out_file.wav";
On 2015/06/18 03:38:15, andrew wrote:
> Prefer not to use non-POD types as static globals. Either make this a const
> char* or move it into void_main.

Done.

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:103:
std::string noise_file_name = noise_name;
On 2015/06/18 03:38:15, andrew wrote:
> You have three variables referring to each of the filenames. I think you can
get
> away with just using the FLAGS variant everywhere.

Done.

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:111:
ASSERT_EQ(stat(in_name, &in_stat), 0) << "Empty speech input.";
On 2015/06/18 03:38:15, andrew wrote:
> Not sure these are really needed. How often will you have any empty file?

Done.

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:123:
in_fpcm = new float[samples];
On 2015/06/18 03:38:15, andrew wrote:
> vector or at least scoped_ptr. Doesn't matter if we leak here of course, but
> let's practice good habits ;)

Done.

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:127:
noise_fpcm = new float[samples];
On 2015/06/18 03:38:15, andrew wrote:
> vector

Done.

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:130:
noise_fpcm[i] = noise_fpcm[i % noise_samples];
On 2015/06/18 03:38:15, andrew wrote:
> This is a bit weird. You're restarting the noise file, which may not be a
great
> idea, depending on the content.
> 
> I think better to crop to the shorter of the two files.

For iid noise it's fine, but I see with real world noise things could get weird.
Done.

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:148: for
(size_t i = 0; i < samples; i += fragment_size) {
On 2015/06/18 03:38:15, andrew wrote:
> I think it's more typical to read directly from the file in chunks, but this
is
> OK.

Acknowledged.

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:156:
WavWriter out_file(out_file_name, FLAGS_sample_rate, kNumChannels);
On 2015/06/18 03:38:15, andrew wrote:
> Indent inside scopes.

Done.

https://codereview.webrtc.org/1182323005/diff/140001/webrtc/modules/audio_pro...
webrtc/modules/audio_processing/intelligibility/intelligibility_proc.cc:165: }
while (FLAGS_repeat);
On 2015/06/18 03:38:15, andrew wrote:
> This probably made sense with the piping directly to aplay. But now that you
> write the whole thing to file then read, you're doing to have a gap between
the
> repeats. I think better to just remove this option.
> 
> Not sure this is worth it, but if you want the same behavior, you could open a
> fifo:
> http://linux.die.net/man/3/mkfifo
> 
> Write to it using WavWriter and read from it using aplay.

Done. Removed. I don't think it's worth it.

[ekm, 2015-06-18 18:51:23]

On 2015/06/18 03:39:22, andrew wrote:
> Oh, I just noticed, why is this CL not closed? Did you reopen it?
> 
> You should start a new CL for your attempt at relanding.

Ok, thanks. Will do.