[Ffmpeg-devel] mpegaudiodec.c and armv5te optimizations

[Benjamin Larsson]

Siarhei Siamashka wrote:

> On 10/5/06, Benjamin Larsson <banan at student.ltu.se> wrote:
>
>> I created some some software that could be used to test audio. Look
>> through the archives. The best solution would be to fix my patches to
>> tiny_psnr.c.
>
>
> Thanks, is it this patch?
> http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2006-April/010279.html
>
> Is see it is at least partially committed already. I'm not sure about
> the part still not in SVN, do I really need it?


Those parts are for my attempt at an ISO rms calculator. It is basicly a
scaled version of a sse metric. PSNR can be used also, but noone defined
how to interpret the metric, so one can only say that one decoder has
better PSNR compared to another.

>
> What I'm having troubles now with is shift value. I tried to encode
> some wav file ripped from CD with lame and decode it with different
> decoders (using mplayer, -ac mad, -ac ffmp3, -ac mp3). They all are
> quite similar to each other (according to tiny_psnr stats), but have a
> major difference from the original file, probably because of shift.


And the reason that mp3 throws away lots of information from the audio.
IMHO a sse metric is useless for this, atleast over the whole file. Use
a "good" mp3 decoder and use that file as reference.

> And I have difficulties with figuring this shift out. I'll try to make
> an automatic shift detection patch for tiny_psnr.c to solve this
> problem.


You can use Octave and it's xcorr function to figure out the shift.


MvH
Benjamin Larsson

-- 
new tiny signature