[Mplayer-users] Bug Report: Audio corruption with recent CVS versions

Arpi arpi at thot.banki.hu
Wed May 30 20:30:39 CEST 2001


> You defined
> #define RESET_AUDIO(audio_fd) ioctl(audio_fd, SNDCTL_DSP_RESET, NULL)
> at the beginning of the file and
> added a call to this in lines 2049 and 2408
>         RESET_AUDIO(audio_fd);
> If this gets called my audio completely 
> corruptes -- the IOCTL within the sound driver
> does not seem to work.
Hmm. The it's really another audio driver bug.
Thanks for note and bug tracking, we'll add it to the FAQ,
and also add a ./configure switch to disable it!

> Can anyone tell me, why they were actually
> introduced?

For very low bitrate audio or big audio buffers (alsa 0.5) the
audio buffer delay was extrembly big (up to 8 secs). So when
you seek in the file, it stays 8 secs until audio buffer gets
empty so audio/video from new position can be decoded.
Audio reset IOCTL deletes everything from the buffer so it can
be used immediatelly! Btw there was some notes in the OSS docs
that re-opening audio device after DSP_RESET is recommended.
I ignored it, because seems to work without it, but then I have
to implement this (at least as an option for buggy drivers)

I'll introduce libao (audio out) with these features after
the 0.18 release (next monday). Then poeple can add alsa and
esd support to it.

A'rpi / Astral & ESP-team

mailto:arpi at thot.banki.hu

Mplayer-users mailing list
Mplayer-users at lists.sourceforge.net

More information about the MPlayer-users mailing list