Help with "lining up" an OST to a film's use of said OST...

[takearushfan]

Hey all,
My latest question has to do with creating a 5.1 that isn't fancy, but rather uses the isolated score in the Surrounds. I'm trying to use the official OST and it obviously won't sound right unless the OST's track(s) are aligned properly with the actual film's soundtrack.
I'm trying to use Audition's beat detector (can't recall exactly what it's called), but it's still difficult to line things up.
Surely someone with editing, mixing and such experience can advise on how to properly align the tracks.
Thanks again in advance
- Justin

[smrpix]

Drop the tracks onto new channel in the multitrack mixer, try to get it close, then nudge back and forth until you hear no echo. The OST version may well be a different edit which won't completely align start to finish.

[takearushfan]

The OST version may well be a different edit which won't completely align start to finish.

Yep, I've noticed that it's not as if I can simply take the beginning of the OST's track and assume it'll align perfectly, so long as I get the starting point right. It'd be nice, but no. There are definitely moments when there's more/less space between notes in the OST than in the film's score.

[smrpix]

Yeah. The requirements for a well-designed motion picture soundtrack are different than a well-mixed standalone piece of music. Looks like you're in for a lot of slicing and slipping. At least you've got a good tool for the job.

[takearushfan]

Yeah, I can already tell that it's a pain in the arse. I appreciate your help though

[takearushfan]

You know, one thing I was wondering about is if it's not *necessarily* (but still quite possible) other factors that may be involved. For instance, I can get a starting point extremely close and all will be well for about 5 seconds, then I notice that the track begins to gradually go out of sync. I haven't tried it again today so I'm not certain, but I believe I'm the one that begins to "fall behind" the soundtrack. Can this have something to do with the need to convert the score from 41000 to 48000, or maybe other factors?

[davexnet]

Other factors. Normally, changing the sample rate wont affect the running time.

This is a difficult task. Having the tools and know-how is one thing, Actually doing it, and having the expertise,
is another thing altogether.

[takearushfan]

Okay. Well, it's annoying but I've had success before with Stephen King's IT. WaSC isn't as easy :/

[smrpix]

As Davexnet said, sample rate usually isn't an issue. Frame rate is a more likely culprit, especially if there was a PAL-NTSC conversion somewhere along the line. It's not out of the question that the music was deliberately sped up or slowed down for the edit. Is your video from a broadcast source? Sometimes tv stations will change the speed of a show to fit a time slot.

[takearushfan]

The video is from a NTSC DVD. What's strange is how it seems to be, like I mentioned above, a gradual "de-sync". I mean, it would make more sense if there were pauses in notes throughout the track, but these are lingering chords that the mixer would have to (seemingly randomly) just say, "Uh... I guess I'll start a slowdown here...? *shrug*".

[intracube]

I guess there's a chance that they did multiple takes/recordings of the music for the film. One version could have been used for the film and another might have ended up on the OST CD.

Do the waveforms look the same on the sections you've managed to line up?

Even if they used the same take, it also makes sense that they'd cut/shorten the film soundtrack to get the running time down.

[takearushfan]

Yep, they do look the same. I'm just not doing something right.
Attached is a less than 30 second sample. I think it sounds well synced, but I'd appreciate your input. The full soundtrack is on the left channel; the isolated score on the right.

[lordsmurf]

Which software?
A lot of cheap audio software does a lot of damage to the sound quality.

Huge difference between freeware or some POS $30ware, and Sound Forge.
Excluding Audacity -- it's good.

[intracube]

The only tool I've got is audacity, and this is just a quick observation from staring at the waveform; the right channel seems consistently behind the left channel. At 7 seconds in, the right track is delayed roughly 35ms. At 18 seconds in, it's roughly the same. At 24 seconds it's delayed by nearly 70ms.

To make things easier to see, I split the L/R channels, normalised both, applied a lowpass filter cutting of at around 90-150Hz. After doing this, the offset between the tracks is visually quite obvious. I'll post some example images this afternoon.

I don't know where you'd go from here as I'm not up to speed on the best commercial software to fix this kind of problem.

TBH, I would've thought an advanced sound editor would be able to line the tracks up by analysing the amplitudes of the waveform and shifting one relative to the other until there's a minimum difference between them. It doesn't seem that difficult/complex (famous last words).

[takearushfan]

Well I have Audacity but much prefer Audition. Maybe Audition 6 has that ability but I'm "stuck"/more comfortable with v3.

[intracube]

Filtering out everything but the low frequencies makes the differences obvious:


It would be really helpful if there was a way to make the edits using these filtered versions as a guide and have the changes automatically applied to the originals - while simultaneously making sure the edits happen at zero-crossing points.

I can't figure out a practical way of doing this yet...

Trying to trim visually with the originals is too difficult.

[takearushfan]

Yep, a major pain. I can see when I'm lagging, then I try to make up for it, only to find that it has to be readjusted again, then again, by varying amounts... ARG!

[takearushfan]

Attached is basically just a joke. That is, it proves how difficult it is to accomplish what we've been discussing. Nonetheless, it's interesting, even if the interesting part comes from how atrocious a mix it is.

[takearushfan]

Okay. Just so I'm sure, in this brief clip are 1 to around 6 seconds what it sounds like when the tracks ARE aligned properly? I'm not even sure what to think anymore :/

[takearushfan]

I think the attached is a fairly decent mix. If you don't mind, please check it out and advise as to what is/isn't good about it.
Much appreciated

[intracube]

Okay. Just so I'm sure, in this brief clip are 1 to around 6 seconds what it sounds like when the tracks ARE aligned properly? I'm not even sure what to think anymore :/

The lack of transiatory sounds like drums/percussion will make misalignment hard to hear. Phase issues will still be noticable - and they're there from around 4 seconds in.

It looks like you've copied and pasted one of the sources to both tracks. From the start to 1.139s the waveforms look identical. After that point, the R waveform is different.

I think the attached is a fairly decent mix. If you don't mind, please check it out and advise as to what is/isn't good about it.
Much appreciated

The music is drowning out the dialogue at the end of the phone conversation (might have been like that in the original score though). I can't hear any timing problems in that section.

Is the original film mono or stereo? If it's the latter, you could try extracting the 'centre' from the stereo mix (only keep sounds that are on both channels) to try and preserve the SFX while eliminating the original score. Then add the OST version over the top. Not sure how well that would work though.

[takearushfan]

The music is drowning out the dialogue at the end of the phone conversation (might have been like that in the original score though). I can't hear any timing problems in that section.

Is the original film mono or stereo? If it's the latter, you could try extracting the 'centre' from the stereo mix (only keep sounds that are on both channels) to try and preserve the SFX while eliminating the original score. Then add the OST version over the top. Not sure how well that would work though.

Do you mean how the low-end begins to drown out the part "Now a squad car's coming over there right now, just get out of that house. Jill? Jill?!" and the the sound of her dropping the phone? If so, I'm glad (in a way) to say that's not me. It was originally mixed that way. I'm not glad that THEY mixed it that way. If they were to remaster the film (fat chance) I'd hope they'd take note of that section. It was a while before I even realized that I was hearing her drop the phone.

Sadly, the film's audio is Mono. This was a very bare-bones release from Sony. I think it at least deserves a Stereo mix. The technique you suggest is what I would've tried, had it been a Stereo track. I mentioned Stephen King's "IT" before and how much more successful I was with that one. Well, IT is a 2.0 Surround track. Extracting, isolating and such yields great results in that one. When A Stranger Calls is a different story

[intracube]

The music is drowning out the dialogue at the end of the phone conversation (might have been like that in the original score though). I can't hear any timing problems in that section.

Do you mean how the low-end begins to drown out the part "Now a squad car's coming over there right now, just get out of that house. Jill? Jill?!" and the the sound of her dropping the phone?

Yes, that's the section.

Sadly, the film's audio is Mono. This was a very bare-bones release from Sony. I think it at least deserves a Stereo mix. The technique you suggest is what I would've tried, had it been a Stereo track. I mentioned Stephen King's "IT" before and how much more successful I was with that one. Well, IT is a 2.0 Surround track. Extracting, isolating and such yields great results in that one. When A Stranger Calls is a different story

I don't know enough about the history of film technology, but maybe WASC was only ever released in mono. From here:
"...until about 1975, when Dolby Stereo was used for the first time in films, most motion pictures—even some from which stereophonic soundtrack albums were made, such as Zeffirelli's Romeo and Juliet—were still released in monaural sound,[18] stereo being reserved almost exclusively for expensive musicals such as West Side Story,[19] My Fair Lady,[20] or Camelot;[21] epics such as Ben-Hur[22] or Cleopatra;[23] or dramas with a strong reliance on sound effects or music"

WASC was made only a few years after that (and on a very tight budget) so an original stereo soundtrack doesn't sound likely.

[takearushfan]

Yep, the IMDB lists it as a Mono mix.

They do however have the ability to do a full 5.1 mix (I think), though they probably wouldn't feel any desire to go that far; too much time, effort and dough. Unless they believe, which even I don't (not by a long shot), that it'll sell a zillion units, they won't bother. I just personally feel that it deserves a Stereo release at the least. I wish Sony would sell the rights to Criterion, Anchor Bay or another studio that might be willing to remaster the audio.

I put "think" in italics above as I'm no expert. However, consider the following. I believe there's enough to do anything from a Stereo to full 5.1 -

1. They obviously have the isolated score.
2. There's a French dub, so the vocals must be isolated.
3. The only thing left are the FX and if music and two different vocal tracks are laid "on top" of them then the FX track must be available too.

Expert-like peeps, does the list above add up the way I think it does, or did I miss something that would hinder the potential for a 5.1?

[takearushfan]

Here's an AC3 of the opening credits. Not as well-aligned as the previous post, but not completely awful either. It required 3-4 edits/mixing points.

[takearushfan]

Okay. For this one you really have to watch while listening. Otherwise you won't know what I was going for. This is a really bad mix, but, at least I tried. If anyone has some advice as to how I could've approached this in a more effective manner, please explain. One thing I noticed is that for some moronic reason I muted times when the other channels SHOULD be heard. I was focusing too much on the SL and wasn't really thinking. Please take a look/listen and advise. Thanks - Justin

[takearushfan]

Alright. I don't think anyone is following this thread anymore, as I've become annoying... though it would be nice if some of you were considerate enough to acknowledge my recent posts.
Here's another one where you can tell what I was going for, but need to watch while listening.

[intracube]

I can only listen to your examples downmixed on a pair of headphones. FWIW:

post #25 attachment sounds ok. Nothing stands out as being wrong.

post #26 attachment has a rough edit when she reaches the top of the stairs and the music stops. The telephone ring only coming from the L channel sounds too extreme. Even if the phone is off to the left, there should be some sound from the R channel as the ring would be reflecting off the walls and floor. You might want to experiment with filtering some of the HF from the right channel in addition to making it quieter (until the view cuts to the shot of the phone).

post #27 attachment: when the cars pull up, there's a second copy of the soundtrack playing at a quieter volume which doesn't sound right.

[takearushfan]

I can only listen to your examples downmixed on a pair of headphones. FWIW:

post #25 attachment sounds ok. Nothing stands out as being wrong.

post #26 attachment has a rough edit when she reaches the top of the stairs and the music stops. The telephone ring only coming from the L channel sounds too extreme. Even if the phone is off to the left, there should be some sound from the R channel as the ring would be reflecting off the walls and floor. You might want to experiment with filtering some of the HF from the right channel in addition to making it quieter (until the view cuts to the shot of the phone).

post #27 attachment: when the cars pull up, there's a second copy of the soundtrack playing at a quieter volume which doesn't sound right.

Disclaimer ... for any that check these out, please note that these are extremely rough. I didn't exactly put days of work into them, heh.

Actually, your first note about listening to a downmix triggered a vital factor I need to keep in mind. Just because all is well in a 5.1 doesn't necessarily mean it'll sound proper in Stereo.

Re: post #25, as I said in my edit update, I think this one was fairly well done. Though it might be off by a slight bit. Thankfully I can't really hear any misalignment if there is one.

Re: post #26, the music seems fairly well aligned; maybe off by a bit but not too drastic. Your listening to a downmix might be a slight issue here. I "targeted" the SL for the ringing. This really is a terrible job but I appreciate your input because at least you know what I was going for. I was wondering how to handle the issue of cutting off the first ring, but it's tough because the score is supposed to linger simultaneously. I was considering perhaps combining cutting off the first ring and overlapping that end portion of the score to make it work a bit better. I'm not sure if I described that very well :/

Re: post#27, I don't know how in the hell I missed the terrible misalignment you're referring to. It's quite obvious to me now. Though I claimed I was only trying to match the score, you'll notice that at roughly 30 seconds I tried to pan Jill's left movement. IMHO, I think it's a decent panning job. Now, I used the chatting between her husband and the babysitter and put it in the SR. Again, you can tell what I was going for and that's what's important.

Thanks for listening and your advice

[takearushfan]

Filtering out everything but the low frequencies makes the differences obvious:

[Attachment 13036 - Click to enlarge]

It would be really helpful if there was a way to make the edits using these filtered versions as a guide and have the changes automatically applied to the originals - while simultaneously making sure the edits happen at zero-crossing points.

I can't figure out a practical way of doing this yet...

Trying to trim visually with the originals is too difficult.

Sorry to revisit this but it's still being annoying, leaving me to wonder what the cause is. I know you said "I can't figure out a practical way of doing this yet...", but do you have any idea, as I said, what might be causing the difficulty in the first place?
The theory of the score using different takes than what's used in the film was something I considered, then I analyzed my butt off and it's very apparent that no, these are not different takes. By process of elimination, the only thing that's left (unless you guys have other factors to consider) is the possibility of the score being sped up/slowed when used in the film. Yet another pain in the arse is how the film's soundtrack and the score have different points of starting/ending detection. Hm, how do I word that better? Er, where you can begin to hear a spot begin to fade in.