2 joined avi files, high beep

[Milcoi]

I joined 2 avi files, when playing the movie (at the switching point "next" avi) you hear a high beep and the volume is down.
After about 10 sec. the volume is back again.

regards,
Milcoi

[therealjoeblow]

-What kind of audio?? (AC3? MP3? CBR or VBR? Sample Rate?)

-How were the original files split (Nandub, VirtualDub, VirtualDubMod, AviMuxGUI???)

If it's AC3 originally split by Nandub, the audio is b0rked because Nandub never split files correctly, and often split AC3 streams in the middle of an AC3 frame.

See my other post on this for more detail:

http://forum.doom9.org/showthread.php?s=&threadid=85114

To fix this high pitched squeal and drop out, you need to demux the audio stream from each file (part 1 and part 2) and then run besplit on each one (get it from dspguru's besweet site) which can fix the broken AC3 frame 9provided of course that you can figure out the commanline parameter - it took me quite a few tries to get it corect, despite the examples in the readme file!)

Once fixed, you can then remux and join the files. But be aware, if the audio duration doesn't match the video duration on part 1 (as noted in my earlier referenced post) - even if you fix the broken AC3 frame, you will get audio desync because the AC3 streams are just appended when joining to AVI, so the 2nd audio stream will be 'pulled up' to where the first one ends, and therefore be out of sync with the 2nd video stream by the difference. As far as I'm told, none of the applications 'pad' the duration that the audio stream is short to make it match the first part's video duration, they all just append the streams leading to the descyn problem when joining to AVI.

I understand that when muxing to MKV this isn't an issue because the MKV specs require audio frames to be muxed with actual timestamps, and therefore the audio for part 2 will always start when the video for part 2 does, regardless of the actual duration of the audio stream in part 1 (I haven't tested this extensively yet, but on limited testing it does seem to be correct so far).