[FFmpeg-user] Using the asyncts filter

Dainius (GreatEmerald) pastas4 at gmail.com
Sat Aug 25 21:44:54 CEST 2012

I have read about the asyncts audio filter and tried it out, but it
seems that the audio file it produced is exactly the same as the input
file (to the point that both had the same md5sum). So my question is,
when exactly is this filter useful? What are the requirements for it
to work? For instance, the description of the filer is "Synchronize
audio data with timestamps". Does that mean that the audio file itself
has to contain timestamps? If so, in what codec and format does it
have to be to have the timestamps recorded?

Also, the arguments for this filter are not very clear to me. What
does the "compensate" argument expect as value, if anything? "1" or
"0"? "yes" or "no"? "true" or "false"? Any of these? And what would be
a sane value to use for the "max_comp" argument?

And finally, this is a bit of a side question, but how do I tell
FFmpeg to only decode a video stream and not re-encode it? In this
case, using the asyncts filter, you need to have a video file to
synchronise the audio to, but the video file is unchanged in the
process. So it would be nice to have a way for the video file to be
decoded, but not re-encoded to anything, saving time. I suppose that
"-vn" could be used here, but when using that, the process completes
so fast that I wonder if the video was decoded to begin with.

More information about the ffmpeg-user mailing list