Timestretching?

[freddo]

Hi,
I have a slightly unique problem for this forum that may be I should be looking at some video forum instead?

I have two copies of a video file, one has english audio, the other is far better quality.
I have extracted the english audio from the inferior video ready to merge with the superior video.
I then truncated the end, but the beginning is timed perfectly already.
But when I merge them the audio drifts out of sync as the video progresses, to the point that there's a great delay between action and speech.

From what I can gather, I'm no expert with video production, as the english audio file is 54 seconds longer than the superior video, I need to timestretch the audio down by 54 seconds.

Is this possible? Any ideas on the relevant software to achieve reduced length using time rather than ratios or percentages?

Thanks for any advice

[whyterabbyt]

Sounds like a sample rate mismatch, caused by importing the english audio at its own sample rate rather than converting it to match the video file's sample rate.

That shouldnt need timestretching as such, I would have thought most video software would have the option to do a sample rate conversion at the import stage. ( If not, basic audio software like Audacity should do it.)

[freddo]

Would that reduce the overall length from 1:35:45 to 1:34:49?
I don't understand how it could be the sample rate? Not that I'm saying you're wrong. Just that to me it's the length of the audio file that's the problem.

[nordickvr]

Sample rate will definitely alter the length.
Anyway, can definitely trust whyterabbyt advice.

[whyterabbyt]

Would that reduce the overall length from 1:35:45 to 1:34:49?

It could do. Although I'd misread the time difference; that's actually a much smaller mismatch than I'd got the impression of. Typically I see 44.1K versus 48K mismatches, which are about a 10% speed/length differences, your case is obviously a lot lower (0.007% if I got it right).

I don't understand how it could be the sample rate? Not that I'm saying you're wrong. Just that to me it's the length of the audio file that's the problem.

The length of an audio file in samples and its sample rate are the two factors which determine how long it takes to play. If you play back at the same rate it was recorded at, it should play back in the exact same duration as it originally did.

If the length of the file isnt changed, then speeding up or slowing down the sample playback rate will make it take less or more time to play. Depending on the software doing the playback, the sample playback rate can be entirely independent of the intended sample playback rate, the one it was recorded at.

Just a thought, but is it possible the two videos are actually different lengths?

For the rather small difference you're talking about, it might be easy enough to split your audio into segments and manually realign them.

[freddo]

Just a thought, but is it possible the two videos are actually different lengths?

For the rather small difference you're talking about, it might be easy enough to split your audio into segments and manually realign them.

Haha! Yes that's exactly the problem, sorry for not stating that more specifically!
Chopping the file would mean a great deal of work! Wouldn't time stretching do the job?

[whyterabbyt]

Just a thought, but is it possible the two videos are actually different lengths?

For the rather small difference you're talking about, it might be easy enough to split your audio into segments and manually realign them.

Haha! Yes that's exactly the problem, sorry for not stating that more specifically!
Chopping the file would mean a great deal of work! Wouldn't time stretching do the job?

It would, but timestretching has the connotation of doing a lot of additional processing in order to preserve the pitch of the original, so its a more sophisticated thing than basic sample rate conversion.

Given the very small difference in length, though, (less than 0.01%), that pitch difference isnt actually going to be particularly noticeable, so simpler resampling would almost certainly be just as good, and probably easier to do.
The free Audacity audio editor has a 'change speed' effect which has enough decimal places of precision to do what you want, I think that should work.

[Etienne1973]

I have the Zynaptiq TimeFactory II software. I can timestretch or (as whyterabbyt suggests) change pitch and speed simultaneously of your file with frame precision.

[freddo]

I did try Audacity but was unsure how to do it. Thanks for the tip.

It worked but the speech is too late for action! The weird thing is that it starts and finishes in sync but the rest of the movie is out of sync??

[freddo]

Thanks for the offer Etienne1973
I think the issue may be more complex than simply shrinking and syncing the audio.
If you think you could sort it though I'd very much appreciate it

[Etienne1973]

It worked but the speech is too late for action! The weird thing is that it starts and finishes in sync but the rest of the movie is out of sync??

I think the issue may be more complex than simply shrinking and syncing the audio.

I don't have the answer. But found the following tutorial:

[freddo]

Thanks for the link
I gave up on this in the end. It's too much hassle as the audio starts in sync goes adrift at some point not long into the video, then drifts back in sync near the end?

Bit strange really!

Thanks for all your help though