Audio precision in lossless and long time to encode.

[Danixu]

Summary of the bug:
Hi, first of all, i'm sorry for my english.
I've tried the last ffmpeg build and i get a problem with Audio precision in lossless audio file, the audio is delayed about 2 seconds.
The audio is passed to another program throught a pipe, but i know that the other program isn't the problem, because with a very old ffmpeg version (SVN 16573, Jan 13 2009) works fine, and i've tried with wav output without pipe and get the same problem.

If you want, here's a test wav files:  http://www.mediafire.com/?1ewceliyjhspedk

The command line that i've used is :
ffmpeg_old -i "input.ape" -ss 00:59:18.71 -t 00:00:15.55 out_oldffmpeg.wav
ffmpeg -i "input.ape" -ss 00:59:18.71 -t 00:00:15.55 out_newffmpeg.wav

How to reproduce:

% ffmpeg -i input -ss 00:45:25 -t 00:05:25 -ac 2 -f wav -

The output is a pipe to another program.

'''ffmpeg version''' N-44080-gdfb4757
'''built on''' Aug 31 2012 16:58:51 with gcc 4.7.1 (GCC)
configuration: --enable-gpl --enable-version3 --disable-pthreads --enable-runt
ime-cpudetect --enable-avisynth --enable-bzlib --enable-frei0r --enable-libass -
-enable-libcelt --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-l
ibfreetype --enable-libgsm --enable-libmp3lame --enable-libnut --enable-libopenj
peg --enable-librtmp --enable-libschroedinger --enable-libspeex --enable-libtheo
ra --enable-libutvideo --enable-libvo-aacenc --enable-libvo-amrwbenc --enable-li
bvorbis --enable-libvpx --enable-libx264 --enable-libxavs --enable-libxvid --ena
ble-zlib

The other thing that i wan to answer is why the program take a long time to convert a portion of a file, for example:
ffmpeg -i input.ape -ss 00:45:15.42 -t 00:00:15.25 output.wav

the program stay a long time without do anything and then start to convert. I take about 2 minutes to convert 15s of audio.

[Cigaes]

Does this happen only with APE files, or does it happen also if you first convert the whole file to WAV or FLAC?

Concerning the second issue you mention, this is completely normal: since you put the -ss option after the -i option, it works by skipping: the input file is decoded and the unwanted samples are dropped. This is slower but more accurate than seeking, which is used when -ss is before -i.

[Danixu]

I think that is only with APE files, because i've tried with a FLAC and works, then i've converted that FLAC to APE and the problem back again. With WAV works fine.

Thanks for the answers.

[Cigaes]

Possible explanation of the problem here:
http://ffmpeg.org/pipermail/ffmpeg-devel/2012-September/130449.html
No fix yet. Thanks for the report.

[Danixu]

Thank you for your attention :)

[Danixu]

Hi again friend, i've seen that if i use -ss xx:xx:xx.xx before -i the position is more accurate than if i put it after -i.

I comment because maybe is useful.

Bye ;)