[FFmpeg-soc] AAC integration

[Michael Niedermayer]

On Mon, Mar 10, 2008 at 11:21:24AM +0100, Andreas Öman wrote:
> Mike Melanson wrote:
> 
> > Just curious, what have you done for testing and validation? Does it
> > "pass" the conformance suite (to within reasonable PSNR)?
> 
> Just conducted some tests, on some streams it seems ok, on others
> not. These are al00_48 to al05_48.
> 
> stddev:  0.70 PSNR:99.34 bytes:716800
> stddev:  0.69 PSNR:99.43 bytes:786432
> stddev:  0.69 PSNR:99.42 bytes:786432
> stddev:  0.69 PSNR:99.43 bytes:786432
> stddev: 19.86 PSNR:70.36 bytes:786432
> stddev: 30.43 PSNR:66.65 bytes:1417216
> 
> So, there seem to be some kind of issue still lingering.
> I've also discovered that it outputs the samples inverted.
> This should be as well fixed I guess.
> 
> I've only tested the last one with FAAD (there seems to be
> some kind of bug in faad/ffmpeg integrations making it always
> output stereo, the first 5 streams are mono)
> 
> stddev:  0.70 PSNR:99.36 bytes:1417216
> 
> Apparently FAAD does the right thing.
> 
> OTOH, if I compare the output from FAAD and aac.c using an AAC
> stream from iTunes (this is what i've been used previously),
> it looks ok:
> 
> stddev:  0.05 PSNR:122.12 bytes:38842368
> 
> Probably, the reference streams is exercising some tool which is
> not commonly used and where it might be some bug lurking.
> I cannot tell any difference by just listening on the streams though.
> 
> Either way, I'll have a look at it and come back when I know
> whats going on ..

Some tool in AAC uses random(), this of course will not match between
decoders. I dont know though if this is the cause of any of the above.

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

I know you won't believe me, but the highest form of Human Excellence is
to question oneself and others. -- Socrates
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-soc/attachments/20080310/e2ef0916/attachment.pgp>