Live recording vs ADR

When filming for movies or TV, sound is invariably recorded live on set. But for various reasons, that audio can be unusable, or new dialogue has to be added in after the fact to help clarify a scene. This is done in a sound booth in a studio, and called ADR, for Additional Dialogue Recording (or Automatic Dialogue Replacement, aka looping).

But the problem is, no matter how hard they try, ADR never sounds the same as live recording. The booth is so clean and soundproofed, it can’t recapture the sound of speaking in the open air or in a particular room, be it large or small.

But why can’t they digitally alter it to match? Surely in this day and age of astonishing digital manipulation they have standard methods to analyse the waveform of live audio and can alter the recorded audio to match.

What’s the difficulty that’s holding them back?

I would guess that most of the time they *do *make it match and it is unnoticeable. It is only when they do it poorly that it stands out.

Exactly. There’s a lot more ADR used than you probably know.

When it is obvious, though, I wonder if it’s because trying to make it fit wound up making the dialog unclear.

You may be making the same mistake crop circle people do: “only true crop circles have broken stalks, fake ones are just bent.” Since no REAL crop circles have yet been certified, everything is just guesses.

Unless you know for sure if ADR was used in a movie or scene, only the worst cases will you notice. You can be sure good sound recording engineers can fool even you.

Recordings made in a soundproof booth (anechoic chamber is more like it) are rarely used without something added; EQ, echo, or other processing.

I say Nay! I don’t believe that I am missing any ADR that matches live recording at all. I would love to be proven wrong, though.

The rule of thumb is that any time you can’t see the lips moving, it’s ADR. If you CAN see the lips moving, it’s only probably ADR.

You’d be surprised how difficult it is to get clean sound recordings on set. There’s so much ambient noise most places it can be almost impossible. Any time they’re shooting on location it’s almost a guarantee there will be some ADR.

Just as a random cite, according to the director’s commentary, Joe Lo Truglio’s character in “I Love You Man” originally had a high-pitched voice throughout the film. They decided it was too over-the-top, so they ADRed it and left some occasional squeakiness, as though his voice was cracking.

Of course, maybe I’m not as sensitive to it as you are, but I hadn’t noticed ADR before hearing the commentary. And I have noticed bad ADR on occasion.

My point is, I can tell exactly when its ADR, and I am annoyed and surprised that it’s so obvious.

OK, I’ll grant you may be able to tell if it’s (bad) ADR. But how can you be so sure, when you don’t think it’s ADR, that it is original recording? Could it be really good ADR?

In theory, yes. In practice, considering it happens in even the biggest and best productions, I doubt it. But it’s a hard thing to know, unless I had some kind of evidence.

I’m afraid I am no longer involved with Hollywood film or music production to provide that evidence, but it sounds like you might not be close to that industry today, either. What makes you so sure your observations trump the professionals?

I seem to be going round in circles. I will clarify.

Why can’t bad ADR have auto settings that make it sound right? What’s stopping it to have a filter that says “out in field” or “in hospital corridor” that actually works, instead of the default “in sound proof booth”.

It’s worth noting that ambient noise is often recorded so that it can be mixed in with the ADR.

You seem to be under the impression that one can press a magic button and make one thing sound like an entirely different thing. That’s not how it works. Sound can be edited and modified, of course, but to do so convincingly takes a talented sound designer and recording engineers, and, especially important, actors who can do convincing ADR performances. (It’s not easy.) If anything goes the slightest bit awry along the pipeline, it can end up sounding like crap.

And if it’s just one scene that only a few people are going to notice sounds like crap, and it’s months later and your actors are off on another production, and your original sound designer died in a motorcycle accident, and the set designs have since been lost, and the original mix settings have been lost to the mists of time, well that scene just might be a bit fucked.

But I also echo what the other posters above say. There’s a huge amount of ADR work that you definitely don’t notice.

Well that’s just unacceptable!

Okay, here’s what I do know, from my own amateur audio manipulation. There are multiple kinds of audio filters that can alter a file. You can change settings manually using these filters, and then you can save various states as a macro to reuse again and again. Any given audio engineer can, in theory have a few thousand different macros stored up for different audio situations.

This is, effectively, “pressing a magic button” only it’s multiple buttons expertly manipulated.

I know this happens in all the other branches of filmmaking that have a digital component, such as visual effects, colour grading, editing, and sound effects. Why not, thinks I, ADR as well?

There are two ways I recognise ADR. One is a slightly misaligned words-to-mouth-movement, which probably happens about a third of the time or better; much less frequently than it used to. And the other is when the sound just isn’t right. It’s flat and clean to the point of glaringly obvious. Or it may be only a subtle difference, but I can still pick it up.

There is a third situation where it is obvious - they’re in a clearly very noisy environment and the talking is a whisper, yet can be heard perfectly. That’s a big clue.

But my question, which has not been addressed at all and instead you’re all focusing on some other aspect, is this:

WHY can there not be filters to fix this problem? Because it’s obvious to me that there is not. Too often ADR is fucking horrendous, and only occasionally is it almost passably acceptable, yet still not right. And we are in a technological era where this should not be happening.

I realise there may not be an answer, and that’s why it hasn’t been answered, but it still annoys me when I ask one question, and people reply with an answer to a different question.

So is your question “why do sound people sometimes make mistakes?” or “why can’t computers perform magic?”. Or, are you asking both?

What’s magic about it? I want to know what’s preventing audio manipulation technology from accurately recreating the atmosphere of real environments.

Well, they *do *use software that automatically aligns ADR tracks with original vocals, and they do use filters to emulate production mic characteristics and ambient recording noise. Since 99% of the time they do it seamlessly and without your awareness, it appears that those sound technicians and their computers and software do a pretty good job. The small number of time where there are glitches, it could be that they had a schedule to keep, that they were lazy, that they just overlooked the issue, that it would be costly to correct properly, or any number of real world events that get in the way of people and machines making perfect products 100% of the time.