philippe_44 wrote:
> - 10 seconds is likely related to the size of the output buffer. The
> error happens at the very last *decoded* bytes and you have always ~10s
> un-played audio in the buffer, likely - that explains regularity
>
> - faad is always implicated when you use a SB3 in the mix as they don't
> do mp4/aac. In that case, LMS does the transcoding to flac
>
> - you can take every file that ends early and try to run it through
> ffmpeg to verify
I'm still not quite understanding. If I have 10 seconds of output
buffer at any given time, then I always have between 0 and 10 seconds of
decoded ready to play audio in that buffer. It fluctuates, because
buffer. I imagine that the buffer can fill much faster than it can
drain, because decoding can happen faster than real time, and because
buffer.
- When I start playing a file, that buffer fills, bits are streaming,
and eventually audio comes out my speakers.
- When the decoder encounters a fatal error, the buffer stops being
replenished and drains, and music stops coming out of my speakers
about 10 seconds later.
- If that fatal error is at, say, 30 seconds into the file, I'd expect
audio to be interrupted at about the 30 second mark.
- If that fatal error is at, say, the last 10 seconds of the file, I'd
expect audio to be interrupted at about the last 10 seconds mark.
That is because I should hear the audio that precedes the fatal error,
and nothing after that point.
I don't understand how a fatal error at the :30 mark of a 2:00 song
would result in 1:50 of the song playing, rather than :30 of the song
playing.
Put another way, if I were having errors with these files at various
time markers, I'd immediately suspect the files. And indeed, I did
until I started seeing that they were all stopping 10 seconds before the
actual end of the song. And if the files exhibited the problem in all
playback circumstances, I'd definitely suspect the files. But I know I
can fully play these files back in iTunes and when sent to a singular
player.
The consistency of when the error occurs (10 seconds before the end of
variable length songs) is notable, as is the fact that when playing
these files to a singular player, I hear the full song to completion.
And given that these are files from different sources and encoded at
different times with different software...it's a lot of coincidence.
I'm not trying to say that the files are NOT incorrectly encoded.
Rather that I don't understand why incorrect encoding results in these
files playing to completion in one circumstance, and not in another.
I'd expect that if the file can be successfully decoded to completion in
one circumstance, it should be able to be successfully decoded to
completion in another circumstance.
I do think I'll run them all through ffmpeg and see what I can glean
from that.
Thanks again for all of your help with this weird situation.
------------------------------------------------------------------------
benh's Profile: http://forums.slimdevices.com/member.php?userid=6732
View this thread: http://forums.slimdevices.com/showthread.php?t=114439
_______________________________________________
plugins mailing list
plugins@lists.slimdevices.com
http://lists.slimdevices.com/mailman/listinfo/plugins
