Syncing Audio From Different Video Sources With Different FPS

[StaticSense]

I did some research, but wasn't able to find exactly what I need help with, so I figured i'd come to the pros and ask for advice.

I'm not having trouble literally muxing audio. I'm having trouble syncing the audio files from two different sources, into a single file.

For instance:

Two video files

Video file 1 - 50.000 FPS - Audio 44.1kHz, 2 channels
Video file 2 - 29.970 FPS - Audio 48.0kHz, 2 channels

I want to take the audio from video file 2, and mux it with the video of file 1.

I've already stripped both files, so now i'm left with video file 1, which now has no audio track, and an mp3 of the audio from video file 2.

I've tried muxing them, and running it through avidemux and time shifting the audio, but it never seems to sync up properly. It'll seem like it's synced at the beginning, but it slowly gets out of sync. I assume that this has to do with the video being at a higher fps than the original, but I could be wrong.

In my mind, because the fps is higher, that would mean that the frames are moving faster in the first video, than in the second, and the audio has been adjusted accordingly. Maybe i'd have to speed up/slow down the actual audio track in order to accomplish this, but again, this is all just a guess, as I'm not really sure.

Hope somebody could help me out with this, thanks!

[hech54]

an mp3 of the audio from video file 2.

What were the properties of this file while still attached to the video?

[manono]

There are just too many possibilities to make this an easy job, or even to provide accurate advice. In addition, there's a good chance the people that 'made' the videos didn't know what they were doing. As a possible way out, compare the audio/video lengths and make one match the other.

[StaticSense]

an mp3 of the audio from video file 2.

What were the properties of this file while still attached to the video?

50.000 FPS
MPEG-4 Visual
AC-3 44.1kHz, 2 channels

29.970 FPS
MPEG-4 Visual (DivX 4)
MP3 48.0kHz, 2 channels

I basically gave you the originals, as I didn't convert anything. Just used toolnix for the video, and another program for the audio.

[StaticSense]

There are just too many possibilities to make this an easy job, or even to provide accurate advice. In addition, there's a good chance the people that 'made' the videos didn't know what they were doing. As a possible way out, compare the audio/video lengths and make one match the other.

there is only 25 seconds difference between them. Would they have to be exact for the sync to work properly? Because i've tried matching them up by time shifting the audio already, and it matches up for a short period of time, but then seem to go out of sync for some reason.

[manono]

25 seconds means nothing unless we also know the total length of the videos. Have you done as I suggested? I'm thinking one's PAL and the other is NTSC, but there's not enough information yet to determine if that's the cause. Are these movies? This synching only works well if the two videos have the same number of frames. Do they?

[StaticSense]

25 seconds means nothing unless we also know the total length of the videos. Have you done as I suggested? I'm thinking one's PAL and the other is NTSC, but there's not enough information yet to determine if that's the cause. Are these movies? This synching only works well if the two videos have the same number of frames. Do they?

They are movies.

File 1 = 50.000FPS
Duration: 1:32:58
(does not indicate PAL or NTSC in either MediaInfo or VLC)

File 2 = 29.970 FPS
Duration: 1:33:23
NTSC


Sorry, I was confused by what you were asking. Were you saying that I should cut the audio file, to make it match the length of the video file? I can do that, no problem.

[manono]

Well, it's partially an NTSC/PAL problem except, if the content was otherwise identical, there should be nearly a 4 minute difference, not 25 seconds. So, I don't guess the frame count is the same for both, which would make it pretty easy to solve. My advice is to forget about it, unless you can find another better source whose audio you can keep.

Were you saying that I should cut the audio file, to make it match the length of the video file?

No, I was suggesting you shorten it to be the same length as the 50fps one. That term is 'stretch' even though, in this case you'd be 'squeezing' it. Not by cutting but by making it shorter in length.

[StaticSense]

Well, it's partially an NTSC/PAL problem except, if the content was otherwise identical, there should be nearly a 4 minute difference, not 25 seconds. So, I don't guess the frame count is the same for both, which would make it pretty easy to solve. My advice is to forget about it, unless you can find another better source whose audio you can keep.

Were you saying that I should cut the audio file, to make it match the length of the video file?

No, I was suggesting you shorten it to be the same length as the 50fps one. That term is 'stretch' even though, in this case you'd be 'squeezing' it. Not by cutting but by making it shorter in length.

I haven't gone through both of them in order to see if maybe there's an extra 25 second scene or something, which could be the issue here, so that's my fault, I definitely should have mentioned that, and now that I think of it, it's probably the reason why it seems to be in sync early on, and not later.

I'm just having trouble with the terminology, I guess. When I say "cut", i'm saying that i'm going to trim 25 seconds off the end of the credits, because that wouldn't be a big deal, and then the audio would be the same duration as the video, and then once I time shift it correctly, it should be in sync, or at least that's what i'm thinking.

I'm still a little confused. In theory, if the videos were the exact same, scene for scene, except one was 50.000 FPS, and the other was 29.970 FPS, and the audio duration were the same, would the sync work perfectly, or would there still be an issue solely because one is NTSC and the other is PAL?

[jagabo]

Video frame rate is independent of audio duration (there are many ways the frame rate can be changed, some change the running time, some don't). The length of the audio and video in an A/V file don't need to be the same. They don't need to start at the same time or end at the same time. Releases in different countries are often different cuts of the film -- with scenes cut or added. If one was recorded off broadcast TV it may have been sped up or cut to fit a time slot. Or someone who recorded a broadcast may have cut out the commercials leaving different amounts of dead time between sections. A bad capture of broadcast TV may have missing frames, changing the running time. There are just a zillion possible reasons for why your two audio files don't sync. You have to figure out what the problem is before you can fix it. The fix may not be as simple as adjusting the initial delay/advance and running time.

[StaticSense]

Video frame rate is independent of audio duration (there are many ways the frame rate can be changed, some change the running time, some don't). The length of the audio and video in an A/V file don't need to be the same. They don't need to start at the same time or end at the same time. Releases in different countries are often different cuts of the film -- with scenes cut or added. If one was recorded off broadcast TV it may have been sped up or cut to fit a time slot. Or someone who recorded a broadcast may have cut out the commercials leaving different amounts of dead time between sections. A bad capture of broadcast TV may have missing frames, changing the running time. There are just a zillion possible reasons for why your two audio files don't sync. You have to figure out what the problem is before you can fix it. The fix may not be as simple as adjusting the initial delay/advance and running time.

Yeah, I see what you're saying. I guess in my mind, I was wondering if it's more of a technically thing, rather than just missing/edited scenes and stuff of that nature. When I think 50 FPS, i think that every second, there are 50 frames, and if another file is only 29 FPS per second, then there are only 29 frames in that second, and because of that, wouldn't it mean that one movie is moving at a faster rate than the other, as if the scenes are sped up by fractions of a second? Or am I completely wrong here?

[jagabo]

When I think 50 FPS, i think that every second, there are 50 frames, and if another file is only 29 FPS per second, then there are only 29 frames in that second, and because of that, wouldn't it mean that one movie is moving at a faster rate than the other

Not necessarily. Let's look at a simpler case to start with, 25 fps vs. 50 fps. If you simply show the 25 fps frames at 50 fps, ie, just flip through them faster, then yes, you will cut the running time in half. But if you duplicate every frame of the 25 fps video you will have twice as many frames -- and displaying the result at 50 fps will give exactly the same the running as the original.

Film has historically been shot at 24 fps. In PAL countries it was usually sped up to 25 fps for broadcast. That led to a 4 percent shorter running time and a 4 percent rise in pitch of the audio. That is known as the "PAL speedup". But there are other ways the 24 fps film can be displayed at 50 fields per second, some of which don't change the running time, some that do. For example, you could simply duplicate every 24 frame, increasing the number of frames by 4 percent so playing them 4 percent faster gives exactly the same running time (and a little jerk every second). Another way is to duplicate at the field level so every 12 frames becomes 25 fields. This also retains the running time but creates a tiny jerk every half second. Yet another way is to create in-between frames that are a blend of adjacent film frames. This method also retains the running time and pitch.

In NTSC countries film was usually slowed to 23.976 fps (about 0.1 percent), then frames were displayed for the duration of 2 or 3 fields, alternating between the two. So half the frames were displayed for 2/59.94 seconds, the other half for 3/59.94 seconds. This retains the running time and pitch (aside from the 0.1 percent change caused by slowing from 24 fps to 23.976 fps). But the difference in display time of the film frames results in a visible "judder", especially in smooth panning shots.

[StaticSense]

When I think 50 FPS, i think that every second, there are 50 frames, and if another file is only 29 FPS per second, then there are only 29 frames in that second, and because of that, wouldn't it mean that one movie is moving at a faster rate than the other

Not necessarily. Let's look at a simpler case to start with, 25 fps vs. 50 fps. If you simply show the 25 fps frames at 50 fps, ie, just flip through them faster, then yes, you will cut the running time in half. But if you duplicate every frame of the 25 fps video you will have twice as many frames -- and displaying the result at 50 fps will give exactly the same the running as the original.

Film has historically been shot at 24 fps. In PAL countries it was usually sped up to 25 fps for broadcast. That led to a 4 percent shorter running time and a 4 percent rise in pitch of the audio. That is known as the "PAL speedup". But there are other ways the 24 fps film can be displayed at 50 fields per second, some of which don't change the running time, some that do. For example, you could simply duplicate every 24 frame, increasing the number of frames by 4 percent so playing them 4 percent faster gives exactly the same running time (and a little jerk every second). Another way is to duplicate at the field level so every 12 frames becomes 25 fields. This also retains the running time but creates a tiny jerk every half second. Yet another way is to create in-between frames that are a blend of adjacent film frames. This method also retains the running time and pitch.

In NTSC countries film was usually slowed to 23.976 fps (about 0.1 percent), then frames were displayed for the duration of 2 or 3 fields, alternating between the two. So half the frames were displayed for 2/59.94 seconds, the other half for 3/59.94 seconds. This retains the running time and pitch (aside from the 0.1 percent change caused by slowing from 24 fps to 23.976 fps). But the difference in display time of the film frames results in a visible "judder", especially in smooth panning shots.

Very informative, thank you.

From what I'm understanding, it seems like it could be possible to convert the frame rate of a file. Would I be able to convert the 50 FPS file, down to 29 FPS? At that point, I would be able to just mux the the audio, and it would be fine. But I assume it can't be that easy, and there are probably huge issues that come along with converting the FPS of a video like that. If this can be done, any idea how to go about it?

[jagabo]

From what I'm understanding, it seems like it could be possible to convert the frame rate of a file. Would I be able to convert the 50 FPS file, down to 29 FPS? At that point, I would be able to just mux the the audio, and it would be fine.

No, slowing the frame rate from 50 fps to 29.97 fps would increase the running time by ~67 percent. The running time would to from ~90 minutes to ~150 minutes. Way too long. Slowing the frame rate by discarding frames would leave you with the same running time. So you would still have the same audio sync problem.

If your audio is about 25 seconds too long over a ~5580 second movie you need to change the frame rate by about 0.45 percent. So from 50 fps to about 49.776 fps.

[StaticSense]

From what I'm understanding, it seems like it could be possible to convert the frame rate of a file. Would I be able to convert the 50 FPS file, down to 29 FPS? At that point, I would be able to just mux the the audio, and it would be fine.

No, slowing the frame rate from 50 fps to 29.97 fps would increase the running time by ~67 percent. The running time would to from ~90 minutes to ~150 minutes. Way too long. Slowing the frame rate by discarding frames would leave you with the same running time. So you would still have the same audio sync problem.

If your audio is about 25 seconds too long over a ~5580 second movie you need to change the frame rate by about 0.45 percent. So from 50 fps to about 49.776 fps.

I see what you're saying. I mentioned earlier that the 25 second difference may be because of an extra scene, or maybe longer credits, so I'm going to skim through both of them, and see if there are any differences.

From what you're telling me, it solely replies on the duration, rather than the specific FPS? Meaning, If I chop the "extra" scene out of the movie, and try muxing it again, it would work? Hard to explain, but what I mean is, despite the two movies having different FPS, are they both technically "moving" at the same pace? and the audio isn't sped up/slowed down to account for the FPS being different? Just trying to break it down into layman's terms, because I don't know much about this topic.

[manono]

From what you're telling me, it solely replies on the duration, rather than the specific FPS? Meaning, If I chop the "extra" scene out of the movie, and try muxing it again, it would work?

Nope, almost certainly not. One's from a PAL source, the other NTSC. The difference is more than a 25 second extra scene. My guess is there's a lot more than that missing from the 50fps one, something on the order of nearly 4 minutes.

And if these are movies, then the people that 'created' them are idiots. PAL movies aren't 50fps and NTSC movies aren't 29.97fps. As I suggested before, find another source.

...despite the two movies having different FPS, are they both technically "moving" at the same pace?

Nope, almost certainly not.

[StaticSense]

From what you're telling me, it solely replies on the duration, rather than the specific FPS? Meaning, If I chop the "extra" scene out of the movie, and try muxing it again, it would work?

Nope, almost certainly not. One's from a PAL source, the other NTSC. The difference is more than a 25 second extra scene. My guess is there's a lot more than that missing from the 50fps one, something on the order of nearly 4 minutes.

And if these are movies, then the people that 'created' them are idiots. PAL movies aren't 50fps and NTSC movies aren't 29.97fps. As I suggested before, find another source.

...despite the two movies having different FPS, are they both technically "moving" at the same pace?

Nope, almost certainly not.

Alright, I think i've got it figured out now then. Appreciate all the help, and I definitely learned something new, thanks!

[Tomycooler]

I did some research, but wasn't able to find exactly what I need help with, so I figured i'd come to the pros and ask for advice.

I'm not having trouble literally muxing audio. I'm having trouble syncing the audio files from two different sources, into a single file.

For instance:

Two video files

Video file 1 - 50.000 FPS - Audio 44.1kHz, 2 channels
Video file 2 - 29.970 FPS - Audio 48.0kHz, 2 channels
ShowBox Mobdro TutuApp
I want to take the audio from video file 2, and mux it with the video of file 1.

I've already stripped both files, so now i'm left with video file 1, which now has no audio track, and an mp3 of the audio from video file 2.

I've tried muxing them, and running it through avidemux and time shifting the audio, but it never seems to sync up properly. It'll seem like it's synced at the beginning, but it slowly gets out of sync. I assume that this has to do with the video being at a higher fps than the original, but I could be wrong.

In my mind, because the fps is higher, that would mean that the frames are moving faster in the first video, than in the second, and the audio has been adjusted accordingly. Maybe i'd have to speed up/slow down the actual audio track in order to accomplish this, but again, this is all just a guess, as I'm not really sure.

Hope somebody could help me out with this, thanks!

In addition, there's a good chance the people that 'made' the videos didn't know what they were doing. As a possible way out, compare the audio/video lengths and make one match the other.

[badyu17]

From what you're telling me, it solely replies on the duration, rather than the specific FPS? Meaning, If I chop the "extra" scene out of the movie, and try muxing it again, it would work?

Nope, almost certainly not. One's from a PAL source, the other NTSC. The difference is more than a 25 second extra scene. My guess is there's a lot more than that missing from the 50fps one, something on the order of nearly 4 minutes.

And if these are movies, then the people that 'created' them are idiots. PAL movies aren't 50fps and NTSC movies aren't 29.97fps. As I suggested before, find another source.

...despite the two movies having different FPS, are they both technically "moving" at the same pace?

Nope, almost certainly not.

They aren't but this may be TV capture from some European 1280x720p50 DVB source. I've seen some samples from those channels and when they show films they always have duplicate frames. Sometimes in irregular pattern.

If video 1 isn't sourced from that then the guy who captured it didn't know what he was doing.