-
Notifications
You must be signed in to change notification settings - Fork 45
Common errors
Sushi is by no means perfect and cannot handle many complicated cases or missformed input. This page outlines some of the most common errors you might encounter.
Most bugs cannot be fixed easily and you'll have to provide the audio files and the script. Because the original WAV files are usually large, you probably want to do one (or both) of the following:
- If ffmpeg is used for demuxing, run the failed Sushi command line with additional
--no-cleanupswitch. Upload the downsampled mono WAV files produced by Sushi. - Reencode the audio to some reasonably high quality lossy audio codec, e.g. AAC.
These steps will ensure that you have relatively small files to upload. Alternatively, if the files are easily and publicly available, you can just send me a link.
Also, don't forget to include the full log. You can output it into a file by appending 2> filename.log to the end of your command line.
This error indicates that Sushi failed while searching for one of the lines in the provided destination WAV file. It most likely means a genuine bug in Sushi and you should report it.
This happens when one of the chapter groups ends up having only linked events. This is usually a bug and you should report it.
You might notice something similar to the following example in the log.
DEBUG: 0:20:12.35-0:20:14.48: shift: 0.901833333333, diff: 0.026121376082
DEBUG: 0:20:14.48-0:20:16.36: shift: 0.901916666667, diff: 0.020110735670
DEBUG: 0:20:16.36-0:20:19.86: shift: 0.901833333333, diff: 0.024271098897
DEBUG: 0:20:24.36-0:20:25.91: shift: 0.901916666667, diff: 0.013290968724
DEBUG: 0:20:42.67-0:20:43.68: shift: -0.126750000000, diff: 0.485115706921
DEBUG: 0:20:43.68-0:20:44.38: shift: -8.949416666667, diff: 0.441348075867
DEBUG: 0:20:44.38-0:20:45.55: shift: -16.157750000000, diff: 0.500385522842
DEBUG: 0:20:45.55-0:20:45.97: shift: -17.004166666667, diff: 0.515159666538
DEBUG: 0:20:45.97-0:20:46.39: shift: -17.367083333333, diff: 0.266188681126
DEBUG: 0:20:46.39-0:20:47.05: shift: -17.888916666667, diff: 0.490459114313
DEBUG: 0:20:47.05-0:20:47.85: shift: -18.519583333333, diff: 0.506170451641
DEBUG: 0:20:47.85-0:20:49.35: shift: -12.798833333333, diff: 0.527408599854
DEBUG: 0:20:49.35-0:20:49.93: shift: -20.807916666667, diff: 0.460786819458
DEBUG: 0:20:49.93-0:20:50.56: shift: -21.327750000000, diff: 0.416900157928
DEBUG: 0:20:50.56-0:20:51.27: shift: -22.056500000000, diff: 0.463233172894
DEBUG: 0:20:51.27-0:20:52.60: shift: -23.109916666667, diff: 0.501647412777
DEBUG: 0:20:52.60-0:20:53.27: shift: -24.087000000000, diff: 0.416000187397
DEBUG: 0:20:53.27-0:20:53.94: shift: -18.495583333333, diff: 0.451422929764
DEBUG: 0:20:53.94-0:20:54.64: shift: -25.426916666667, diff: 0.542516529560
DEBUG: 0:20:54.64-0:20:56.23: shift: -26.871000000000, diff: 0.498153328896
DEBUG: 0:20:56.23-0:20:56.90: shift: -27.713000000000, diff: 0.495495349169
This usually means that Sushi failed to locate appropriate audio segments for a few consecutive events and the error recovery algorithm didn't work.
If you're sure the appropriate audio segment does exist in the destination audio, try increasing the --window or --max-window parameter. If that doesn't work please file a bug report.