[Ffmpeg-devel] [BUG] garbled sound when playing 8kHz MP3s

[Michael Niedermayer]

Hi

On Wed, Nov 15, 2006 at 11:33:35PM +0100, Stefan Gehrer wrote:
> Stefan Gehrer wrote:
> >>
> >>is there any decoder besides lame which can decode it? if no then i 
> >>would
> >>guess that the bug is in lame though thats just a guess (1 buggy 
> >>program vs.
> >>2 buggy programs)
> >>
> >>[...]
> >>  
> >Actually yes, there is. It is just not publicly available. But I 
> >recently had to test
> >an embedded decoder solution and compared it with open source decoders.
> >The decoder under test and lame succeeded, ffplay/mplayer/mpg123 failed
> >on the (lame-encoded) 8kHz sample. And I can assure you that the decoder
> >under test did not have any code common with lame.
> >I also came across this recently with a file from the Internet, I will 
> >see if I can
> >dig out a link. Of course I would not know if this was lame-encoded or 
> >not.
> >
> You can find some 8kHz mp3s with described behaviour here:
> http://www.addicks.net/astcrapper2k/
> Actually some of them are fun to listen to (for those who understand 
> German):
> That guy tries to trap spam calls on the phone using his Asterisk-based
> answering machine and records the conversations.

unless someone provides either a 8khz mp3 from the reference encoder
or a copy of the spec describing 8khz decoding i wont look into this
any further, also note that all decoders (mpg123, ffmp3 and mad) 
contain notes in the source that 8khz is (likely) broken

from libmad:
/* the 8000 Hz short block scalefactor bands do not break after
   the first 36 frequency lines, so this is probably wrong */
...
 /* this is probably wrong for 8000 Hz short/mixed blocks */
...
/* this is probably wrong for 8000 Hz mixed blocks */

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

In the past you could go to a library and read, borrow or copy any book
Today you'd get arrested for mere telling someone where the library is