If you want to do a shot in slow motion but a person is speaking, should the pitch of their voice be changed as well? This of course can sound cheesy. But if you don't change the pitch is it weird to have a normal dialog pitch if the video is in slow motion?

One movie I can think of that has slow motion is Parasite, in this clip at 2:25 into the clip:

Parasite - Dasong’s Birthday Party (HD) (SPOILERS) - YouTube

But how do they do that without it coming of as strange that the pitch hasnt' changed?

You don't need to use the original sound, just get the actor to say the dialogue slowly in post and replace the original audio. Film making is about cheating.

Oh sure I can do that. I was just wondering how it should sound to the audience, pitch wise, in terms of what would come off as natural.

If it sounds and looks right to you, the audience should follow. unless you're totally lacking in judgment.

I've never seen that movie and my God, is it really that bad all the way through? Seriously - Brian is correct as usual - you create what you need to make it work. I doubt much of that scene has recorded audio from the recording - it sounds fake, and quite averagely done too. If you halve the speed, that's not exactly very slow motion, but the impact on audio is severe, so it's a re-record. Remember Alvin and the Chipmunks - that was a doubling when using the real performers and a bit less with female performers as in the later stuff. half is pretty average slo-mo isn't it!

Oh okay, then thanks, I can do it that way. Thanks.

How slow is that slow motion then in terms of fps?

As for if the movie is that bad all the way through, it depends on how you define bad, I guess.

I find that audiences prefer when I adjust the pitch back to normal in slow motion shots - especially if recognizable voices are involved.

However, the pitch adjustment tools in most NLEs aren't that great. The sound often takes on a harsh, metallic quality. In particular, I find the pitch correction in Premiere to be just barely OK for technical/instructional/ENG types of footage, but it would be completely unacceptable for a feature film. (Most of my slomo is at 1/2 to 1/4 real speed.) You may have to experiment with a few audio editors to find one that doesn't have unpleasant and unrealistic artifacts.

Or, as has been suggested, treat it as ADR and record later.

Oh okay thanks. Yes, I have played with the pitchshifter in Premiere Pro as well and also find it has a metallic quality too it. It's just whenever I have done ADR in the past, it could never match so I am reluctant, unless it's a last resort, or at least need to figure out how match it better.

Ha! I never even knew premiere could pitch shift! I always use cubase for this, as for the past few versions it's pitch shift ability has been really good.

The example you've given would be pretty simple to match up.

A number of feature films have a lot of ADR and get away with it. I've done it manually editing on 16mm film and it works to a satisfactory level. .

It depends on what you mean by match. I know some people obsess about precisely matching waveforms when syncing up sound and picture when using the camera audio as a sync reference with an external recorder. When, in practice, all you need is the waveforms to be within the same frame for things to be in sync. Also, you don't use the camera audio, since it's poor quality.

Oh okay, it's just in the past, I have had trouble trying to match with ADR. For example in this clip, I have had directing the actors to match it, or maybe I just need to develop my directing skills with that? It happens at 9:28 into the video:

Timewine H.264 copy - YouTube

When I did it the old fashioned way, I was replacing the original actor's voice and a couple of words. I played back the original audio to the person re-voicing through headphones.and repeated this until it was close enough. This was called looping in the past for a good reason.

It was fine tuned during the edit by removing the odd frame.

Cubase, my favourite for music has the ability to take a clip - and then take different versions of the same part and then adjust it to fit. Musically, lets say you have 5 BVs and they should all be singing exactly the same, but you discover they all have slight differences - maybe in the gaps, or the phrasing. You designate one as master, and then the others get stretched and shrunk to fit. You can take the compromised audio shot on location, designate it as master, and then the replacement dialogue you recorded at the replacement session gets wrangled to fit. As an ADR tool, it's great. It needs a clearly defined audio file, but doesn't seem to worry too much about background noises as long as the peaks in the voice are distinct.

When I did it the old fashioned way, I was replacing the original actor's voice and a couple of words. I played back the original audio to the person re-voicing through headphones.and repeated this until it was close enough. This was called looping in the past for a good reason.

It was fine tuned during the edit by removing the odd frame.

When you say you fine tune it by removing the odd frame, wouldn't the removed frame be visible in the editing because you would see that there is a missing frame?

Cubase, my favourite for music has the ability to take a clip - and then take different versions of the same part and then adjust it to fit. Musically, lets say you have 5 BVs and they should all be singing exactly the same, but you discover they all have slight differences - maybe in the gaps, or the phrasing. You designate one as master, and then the others get stretched and shrunk to fit. You can take the compromised audio shot on location, designate it as master, and then the replacement dialogue you recorded at the replacement session gets wrangled to fit. As an ADR tool, it's great. It needs a clearly defined audio file, but doesn't seem to worry too much about background noises as long as the peaks in the voice are distinct.

When you say different versions of the same part, do you mean different 'takes'?

Don't you think before you reply? The odd frame in the audio.

Oh okay it's just I've never heard the word 'frame' used in audio and it's always been a video term. People alawys call them 'samples' in audio, but I see what you mean.

However, if you cut out an audio frame though, you can still hear that a frame is missing though, so how do you make it so you cannot hear that?

You delete the audio frame(s), so that the timing between the words changes. You insert frames (from elsewhere on the re-voice recording) if you want to lengthen the gap, but in this case it just involved shortening, You usually have atmosphere tracks in the final mix, which covers any variations, since the recording is done in a studio.

When in the NLE timeline, there are frames in the time ruler. similar to timecode. for all tracks so you can keep sync as you edit. However, I was referring to editing 16mm film, which has magnetic film which looks the same as the photographic 16mm, but it has a magnetic coating. The principle remains the same, just you're physically cutting out the audio frames and splicing it back together.

Oh okay I see. But whenever I cut out an audio frame, you can still hear that a frame has been cut though.

Only if you do it badly, Ryan. It depends on how adept you are with software. Frames are only talked about in audio circles when you are editing audio for video, because audio editors work down to individual samples and video doesn't. If you can only cut your pictures in the video frame gaps, it make sense to work in frames in your audio. Most sync problems are OK to a video frame adjustment as the smallest one, but your audio editor could snip out half a frame if you were really picky.

Audio edits do t work with cuts between dissimilar waveforms. There's a jump or worse, a pop or click. But the secret are cross faces. Problem edits just get a short cross fade and that cures it. As I've said, I use cubase, but for simple edits I'll go from premiere to audition because of the linking, sort it then go back. I do t like audition but it's perfectly capable. Now, I tend to do complicate audio in cubase, other audio work in sound forge and just a few things in audition. I could probably do them all in audition really, but I'm quicker in cubase.

You should check your NLE's audio cut settings, so that you have a softer cut, if it's really obvious. If the audio is the same on either side of the "cut out" section, you shouldn't hear anything. Note that this means not having a gap in your audio track. You can put in a bit of "silence" if there is a gap.

In the days of film editing, the splicer had an angle for cutting the magnetic film, so there was a very slight overlap at the splice. When using a razor blade to edit 1/4" audio tape, the cutting block had a 45 degree cut, to do the same thing. Basically softening the cut with a mechanical cross fade.

You can only do it if there's a gap in the speech and there's clean background sound on the recording.

Your atmosphere tracks etc, should cover things up in the final mix.

Using a NLE I would tend to use trimming with slide if i wanted to adjust by a one or two frames. Just use the razor/slice to cut the audio track, so the previous section isn't changed and slide the audio for the best lip match. Slide may get you a better overall match in the clip, without getting down to word by word lip matching.

Off topic a bit - I did something very silly when I was shooting in a big church with 4 cameras. I used the audio from the camera doing the centre line closeups and the sync was very difficult to match, until I realised the centre camera was the furthest one away, but with the tightest lens - so it was recording sound with a delay so by using that one as the sync 'master' - perfect sync was actually wrong. The correct sync position was ahead of where the waveform from that camera was. I took ages to click why it didn't match!

Oh okay. Yes I can do crossfades to get rid of the popping sounds, but even then, a lot of times you can tell that there is something wrong in the sound, but it's hard to descrive. A tonal shift in the voice sound sometimes.

Another thing I could do is cut to a reaction shot, if I have trouble matching up the dialogue. However, what if a reaction shot would be emotionally incorrect at that moment? What is worse, ADR that may not line up, or an emotionally incorrect cut to a reaction shot?

You shouldn't be changing the voice, just matching the words to the lips.

Oh yes, it's just what I mean is let's say I cut out an audio frame. You may here a change as the frame is skipped over because the voice has changed, if that makes sense?

You don't cut out frames where there is a change in the audio. You only take out frames that have no speech.

This is something that you should practice, since it's a skill.

Oh okay, sure. Taking out frames where there is no dialogue is not a problem, and I can do that easily. But what if you need to make a sentence be shorter to match in ADR, but the actor did not pause at all while speaking?

You'll have to use a different technique, Some NLE's will allow you to fit clips into a certain length within the timeline. However, you're probably best using a DAW for maximum options for this type of stuff.

Well, in Cubase, you just stretch and shrink it to fit.

There are YouTube videos that explain how to do ADR with Premiere Pro and Audition . It should align automatically if you're using the appropriate software. The clue is in the name Automated Dialogue Replacement, using other methods isn't ADR, but fit into the generic term of dubbing.

The method I'm describing is doing it manually, which is a slow business,

Oh okay, thanks, I suppose in the past, I wasn't doing it in an automated way, but it had to be done manually.

In the videos I saw, the automated workflow gave precise lip synchronisation, even when another voice was used. Performance is entirely another matter, that's up to the skill of the actor.