GNOME Bugzilla – Bug 731601
audioconvert: Does not always negotiate the best formats
Last modified: 2014-06-13 06:28:43 UTC
Created attachment 278384 [details] Original pipeline log as run on BSD I've crafted the following pipeline from a situation that was described to me that is actually happening on BSD. I think I've seen this problem on Windows too without realizing that the bad audio quality was due to low depth being chosen. gst-launch-1.0 audiotestsrc -v num-buffers=1 ! audio/x-raw,format=F32LE ! audioconvert name=c1 ! "audio/x-raw,layout=interleaved,rate=44100,format=S8,channels=[1,2];audio/x-raw,layout=interleaved,rate=44100,format=S16LE,channels=[1,2]" ! fakesink In this case, audioconvert will fixate to S8, where S16LE would have given better results. I am not sure why the caps get split this way, but I think audioconvert should try and preserve quality in all cases. Original pipeline was (trace attached): gst-launch-1.0 filesrc location="file.ogg" ! oggdemux ! vorbisdec ! audioconvert ! audioresample ! osssink
Note that reversing audioresample and audioconvert does work: gst-launch-1.0 filesrc location="file.ogg" ! oggdemux ! vorbisdec ! audioresample ! audioconvert ! osssink
Thanks for the bug report. This particular bug has already been reported into our bug tracking system, but please feel free to report any further bugs you find. *** This bug has been marked as a duplicate of bug 706884 ***