[MPlayer-dev-eng] RE: sound being fastforwarded 3 times as fast as the fastfTue Nov 6 11:04:06 2001

[Arpi]

Hi,

> />/ Type: auds   FCC:  (0)
> />/ Flags: 0
> />/ Priority: 0   Language: 0
> />/ InitialFrames: 1
> />/ Rate: 48000/1152 = 41.667
> /  ^^^^^^^^^^^^^^^^^ it is vbr mp3, ok
> 
> >/ Start: 0   Len: 72759
> />/ Suggested BufferSize: 6912
> />/ Quality -1
> />/ Sample size: 1152
> /               ^^^^-- should be 0 to work...
> 
> this file is simply broken. very broken. hack with hexeditor...
> 
> -> i couldn't do the hexfix in the file since it was on readonly in my
> case but I forced the value in the code to zero and it works :-)
> 
> just 2 small additional questions : 
> 
> 1)  The rate line in the header tells you its vbr mp3, while
>     the check in the code is on the sample size. Would it be a correct fix to 
>     check for this rate iso the samplesize in the code ?

it was the first (and i hope the last) file with this problem...
i don't want to add such workaround to the code.

> 2) Any known header validation software or intents to implement this kind of logic 
>    in the player.. at least some major warnings about header inconsitency before  
it is not so simple. detect that header is broken is very hard, i think it's
impossible. the problem is usually bad interleaving or bad avgbps values.
it can't be detected without decoding the whole file, counting uncompressed
samples, rebuild header etc. mencoder can do that if you need.


A'rpi / Astral & ESP-team

--
mailto:arpi at thot.banki.hu
http://esp-team.scene.hu