I have audio I extracted from a 25fps PAL DVD that I would like to supplant onto a video at 29.97fps. The audio is from an alternate language of the film. I can tell the speeds are out of sync because simply trying to overlay the 25fps sound onto the video in iMovie does not work. Can anyone guide me through the simplest and easiest way to "stretch" the sound out correctly? Currently the audio in the language I want is playing a bit too fast when overlayed onto the video.
(This is a cross-post. I first tried the audio forum but they nearly bit my head off and told me to come here, which I usually do.)
+ Reply to Thread
Results 1 to 23 of 23
-
-
Depends what software you have. I'd simply load the audio file into Adobe Audition and stretch it by a factor of 1.1988. If you don't have Audition, maybe you can find a copy of CoolEdit which is what Audition was called before Adobe bought it. There's probably a few shareware audio programmes that can do it too.
Edited to add that I didn't notice this was in the Mac section so ignore the above as the software is PC only as far as I know. You got some duff information in the audio forum by mentioning film and immediately being told it would be 23.97 fps rather than 29.97 fps. If the video you have is from a DVD, it will be 29.97 fps. -
By film, I meant movie--not 35mm. I have Amadeus Pro which has a change pitch and speed feature that I have found to be extremely useful in the past, since you can change pitch without changing the speed and vice versa.
I was just hoping there was some program specifically designed for this that would correctly change the audio to the correct length for 29.97fps without me having to manually tinker around until I got a close enough approximation. -
You don't know what you're talking about. First, his 29.97fps video isn't from a DVD I don't believe or he would have said so (as he did about the PAL source). Second, as I stated clearly, films aren't 29.97fps (and if it were from a DVD, the 'base' or 'real' framerate would still be 23.976fps, and you'd use that as a basis for stretching the audio). Third, your recommendation about stretching the audio by a factor of 1.1988 is so completely wrong that it boggles the mind.
I'll chalk this up to you being from a PAL country and knowing nothing about how NTSC DVDs work. -
Bollocks! The OP says that the audio plays fast when overlaid onto the video, therefore it is being played at a higher, not lower, frame rate than it was recorded on. If it started out at 25 fps and was played at 23.97 fps, the audio would be slower not faster. He even said he needs to 'stretch' the audio to make it fit FFS!
So the audio file needs to be made longer by the correct amount. The correct amount being 29.97/25 = 1.1988 so in the software you have you enter 119.88% in the speed box and leave the pitch box at 100%. This will then make the audio file longer but not lower the pitch, and slow it down when you overlay it on the video. -
Another method is to get the actual length of the audio in question, and the length of the video it is to be matched with.
By opening each in an editor, or even mediainfo, the respective durations can be ascertained.
Once you have the duration of the video, you can stretch the audio to match it. Don't even need to do any calculations.
The free "Audacity" audio editor does a good job at this.
PAL movies , usually sped up 4%, has to have the equivalent adjustment made to the audio, for example the PAL audio is slowed down to fit the NTSC version of a move.
However, it's entirely possible that a 29.97 piece of video and a 25 fps piece of video have the same run times,
and no audio adjustment is required..
Lastly, as was mentioned before, the 29.97 video is really 23.976 with repeated frames, and the adjustment is
only the 4% mentioned earlier... -
Nice try. We'll use your own (faulty) line of reasoning for this.
The OP says that the audio plays fast when overlaid onto the video
He even said he needs to 'stretch' the audio to make it fit FFS!
So the audio file needs to be made longer by the correct amount.
Say you have a 100 minute long movie playing at 25fps. If you slow it to 23.976fps:
100 x 25/23.976 = 104.27 minutes. PAL audio has to be slowed down to match.
If you speed it up to 29.97fps:
100 x 25/29.97 = 83.42 minutes. PAL audio has to be speeded up to match.
Audio from a 25fps source plays faster than that from a 23.976fps source, and ends before the movie. But audio from a 25fps source played with a 29.97fps source plays much more slowly on the video and ends way after the movie.
Have you never heard of 'PAL Speedup" where PAL movies are speeded up as compared to the 24fps it's shown in a theater? Have you never heard of '3:2 Pulldown' where NTSC movies are encoded as progressive 23.976fps but output at 29.97fps, and at exactly the same length as the 23.976fps progressive source? -
Your explanation is perfectly correct when dealing solely with video but I think in this case you are looking at it the wrong way round. I am assuming he has a video that is at 29.97 fps and an audio stream that is at 25 fps. When the two are played together, the audio runs faster than the video because the 25 frames, which should last for 1 second, are actually being played at 29.97 fps so last less than a second (25/29.97 of a second). He's playing something that should be played at 25 fps at 29.97 fps so it ends sooner and isn't in sync with the video because that should be played at 29.97 fps and is running at the correct speed. It's like having two copies of the same 33 rpm record (assuming you are old enough to remember records), playing one at 33 rpm, the other at 45 rpm and wondering why one finishes before the other.
If he stretches the audio by a factor of 119.88% it will last the same amount of time as the video so remain in sync. -
"Tinkering" is the way to go. You'd be very lucky if it was a simple speed change and nothing else. Probably an offset at least.
What I do use use a sound editor like Audacity, which is available on Mac.
Get the soundtrack from the target video.
Get the new soundtrack.
Load them both.
Find a sharp sound effect in both tracks (that you can pinpoint in the waveform) near the beginning, move the new track to sync that point.
Mark that point.
Go to the end, find another sharp sound in each track.
Select the interval between the sounds in the original, let's say it's 1000 seconds.
Now measure the interval in the new track, say that's 950 seconds.
In this case, you need to slow down the new audio by 50/1000 = -5.00%
You can use Audacity's "Change tempo" (without changing tone) or just "Change speed".
Select the entire track and apply.
You may need to move the first sync point a bit after this again. Now you should see both sync points aligned.
But the test is whether all the intermediate ones are too, if the videos are edited differently you will have to resync each scene individually, which will take quite a while.
When it's good enough, mute the original track and export your new track. -
No. Audio has no frame rate, only a duration (technically, a number of samples divided by the sample rate). It is played the same duration no matter what the frame rate the accompanying video is. The OP needs to slow the audio from his 25 fps PAL rip to match a 23.976 fps NTSC rip. That's a ~4.27 percent slow down, ie, increase the length of the audio by factor of 25 * 1001 / 24000.
Don't be surprised if the two releases are different cuts of the movie. In which case no simple or stretch of offset will make the two match.Last edited by jagabo; 24th Nov 2012 at 23:28.
-
Since you read and quoted from the other thread:
https://forum.videohelp.com/threads/351072-Using-audio-from-25fps-source-on-29-97fps-video
did you miss the part where he said it's a movie? Movies, by definition, aren't 29.97fps. NTSC DVDs are most often encoded as progressive 23.976fps, with pulldown applied to output interlaced 29.97fps. The length remains the same and is about 4% longer than the PAL length. If some idiot takes a DVD movie and reencodes it with the duplicate fields and/or frames, keeping it at 29.97fps, the length is still the same as the 23.976fps source.
If he stretches the audio by a factor of 119.88% it will last the same amount of time as the video so remain in sync.
I wouldn't mind seeing 10 seconds of this '29.97fps' video, a section with steady movement. -
No he didn't, he never mentions the word movie, he says film. That is what made me originally think he was in Europe (as he doesn't give his location) as we refer to something produced by Hollywood (or any other production company but intended to be shown in a cinema) as a film, you refer to it as a movie. We have a TV channel called Film4 which shows films, we also have Sky Movies which also shows, what we call, films, as opposed to documentaries, sit-coms, game shows, etc. Anything that isn't a Hollywood 'film' would be referred to as a programme.
As jagabo says, audio doesn't have a frame rate as such, only a duration. The problem here is the duration is wrong as it is being played at the wrong speed. The fact that it is speeded up makes me think it is being played at 29.97/25 so would need to be stretched by 119.88%, if it is being played at 25/23.976 it needs to be stretched by 104.271%. Only the OP knows which is correct as he has the originals. Try one, if it's wrong, try the other. -
Film, movie, it's the same thing. It's something that plays at 24fps in a movie theater and 23.976fps on an NTSC DVD (or 29.97fps with pulldown).
Geez, have you learned nothing from this discussion? If you overlay the audio from this 25fps 'film' over the same 'film' playing at 29.97fps, it plays way too slowly. Which makes no sense anyway because 'PAL speedup' tells you it should play too fast. Which it does when played over the same 'film' running at 23.976fps. And you say you want to 'stretch' it by 119.88% to make it even longer, to play even more slowly? Think, man, think! Did you notice you switched the '25' from the bottom to the top when doing your division for the 23.976fps? And accidentally got that one correct? I already showed you the math in an earlier post and you couldn't even get it right with it staring you in the face.
One would think that when challenged you'd run off and do some research to get your facts straight. Either you did nothing at all, or you didn't understand what you read. I'm not the only one trying to tell you everything you've said in this thread has been dead wrong. Talk about 'duff'! There's nothing wrong with being ignorant if one shows a willingness to learn. We all had to learn this stuff once upon a time. There's something very wrong with being obstinate and obdurate in the face of overwhelming evidence to the contrary. So that I don't keep repeating myself in the face of your continuing foolishness, I'll retire unless and until headless chicken returns with a sample or some results from trying out the solutions offered. -
Originally Posted by manono
Why would I want to do some research on an inferior standard that I have no interest in or will ever use? Quite why NTSC runs at 23.976 fps and is then bodged about with to run at 29.97 fps is irrelevant and of no interest to me. The OP has a video that runs at one speed and an audio file that runs at a different speed. The audio file runs faster so has a shorter duration. The duration of them is different, that is all. The reason why is immaterial, he wants them to run at the same speed so needs them to have the same duration. Therefore, the audio needs to be made longer, it needs to be stretched. He has software capable of doing this but it is the amount it needs to be stretched by that is important. By doing some basic maths it is possible to work out two possible amounts by which the audio has to be stretched, an understanding of where the numbers come from isn't required. It may be that neither of them are correct for whatever reason but they are a reasonable place to start.
Oddly, davexnet, jagobo and AlanHK all seem to have grasped what the problem is and appear to be agreeing with my thoughts, you're the only one trying to make it far more complex than it actually is. -
The method I mentioned, actually finding out the length of the video and speeding up the audio to match is a good place
to start. Then you don't need a formula, just work with the absolute lengths.
I know 29.97/25 is almost 1.2, but you'll never see such a speed up/slow down - it would look ridiculous.
It's not how it's done. -
I apologize for temporarily abandoning my thread. Its been a busy past few days with no time for AV editing.
The other video source is from a NTSC DVD. Sorry for not specifying.
Stretching to ~120% was not the correct amount. It turned out to be in the vicinity of 104.9%, which was determined by setting the output audio length to match the video length to the accuracy of a tenth of a second. Because the NTSC and PAL DVDs are of the same film, but the PAL DVD is an alternate language version with many scenes removed, I cannot simply apply the necessary audio modifications to the extracted track.
As per davexnet's suggestion, stretching to approximately 104.5 worked for a couple of the clips, but I have another clip where simply setting the output audio length to match the video length is not working. It's a real head scratcher. -
The audio and video lengths don't have to match. The audio can start and or end before or after the video.
-
This is the big "Gotcha" . There will always be cases where it wont work; either the video or audio may have something
extra at the beginning and/or end that throws the method out of the window. At least it's a good check and it gives
you an estimate as to whether the approx plus or minus 4% is in the ball park.
I've done some like this in the past; the usual way to proceed in a "difficult" situation is to do first attempt based on your
observations, then align the audio at the first place you can (where you can see some lip synch).
Then skip to the end and see how far out it is and readjust accordingly.
I even tried to do it once, scratching my head, I couldn't get it right - I finally realized I had two different cuts of
the movie; the video had an extra scene that was entirely missing in the audio from the other audio
I was trying to synch. -
-
Normally, when converting PAL to NTSC, as I mentioned earlier, you stretch the PAL audio by a factor of 104.27%. You said both that there were scenes missing from the PAL version, as compared to the NTSC version, and also that stretching it by about 104.9% fixed it. But, of course it didn't really completely fix it. You still have sections where it's noticeably out of synch.
If you really want to 'fix' it, you have to first stretch it by 104.27% and then add silence at all of the missing scenes the same length of those missing scenes as found on the NTSC version. Then if you're lucky, all that will have to be done afterwards is maybe to fix a delay. In cases such as this sometimes AlanHK's method is the way to go. It's a lot of work, a lot of trial and error, and maybe not worth it for you.
This is all assuming the NTSC DVD is progressive 23.976fps with pulldown applied. If it's field-blended from a PAL source and the differing lengths are entirely because of the missing scenes, then that 104.27% stuff doesn't apply. That's one reason I suggested uploading a sample earlier. -
-
-
Similar Threads
-
Using audio from 25fps source on 29.97fps video
By headless chicken in forum AudioReplies: 5Last Post: 26th Nov 2012, 06:14 -
29.97fps subs to a 25fps Video?
By solrage in forum Authoring (DVD)Replies: 34Last Post: 1st Feb 2011, 21:16 -
converting PAL (25fps) to NTSC(23.97fps) with audio
By drew24 in forum MacReplies: 6Last Post: 21st Dec 2009, 09:45 -
Help with HD @ 29.97fps to DVD at 25fps
By Killer3737 in forum Video ConversionReplies: 9Last Post: 19th Nov 2009, 14:29 -
avi is 23.97fps,srt is 25fps
By Soixante in forum SubtitleReplies: 7Last Post: 3rd Dec 2008, 14:26