Fixed-point mad decoder broken

I have not thoroughly checked, but it creates noise, while avcodec works just fine. This may be a filtering/conversion issue though.

My best guess is that libmad assumes FI32 mixing, but we can now use S16N mixing... This may also affect liba52 and libdts (not tested).

added Component::Audio: filters Severity::normal Type::bug Version::master git labels

changed milestone to %2.0.0 bugs

changed title from Fixed point mad decoder broken to Fixed-point mad decoder broken

assigned to @Courmisch and unassigned @MigrationBot

added Component::Core: Audio output label and removed Component::Audio: filters label

commit 8cfc8c96 Author: Rémi Denis-Courmont remi@remlab.net Date: Wed Aug 3 17:23:39 2011 +0300

aout: support indirect sample format conversion (fix [#5150](https://code.videolan.org/videolan/vlc/-/issues/5150))

Original author: remi@remlab.net

added Status::fixed label

closed

commit 5593848a Author: Rémi Denis-Courmont remi@remlab.net Date: Wed Aug 3 16:53:03 2011 +0300

mad: convert as requested by caller (refs [#5150](https://code.videolan.org/videolan/vlc/-/issues/5150))

Original author: remi@remlab.net

commit c54022a6 Author: Rémi Denis-Courmont remi@remlab.net Date: Wed Aug 3 16:45:34 2011 +0300

dca: do not override output format (refs [#5150](https://code.videolan.org/videolan/vlc/-/issues/5150))

Original author: remi@remlab.net