ADR... as in really 'automatic'...

J

j1clark@ucsd.edu

Veteran
I got a news item on 'computer' generated voice for Roger Ebert. He apparently has lost all ability to speak, due to cancer, and the need to remove most of the vocal mechanism and his lower jaw.

What interested me was that the 'voice' that is demonstrated in the URL below, was 'reconstructed' by processing the myriad of recordings of his voice, then packaged, such that when his typewritten material is 'played', his vocalization is used.

While one can definitely hear artifacts of the computer generated voice, it seems far more 'natural' than other CGV output.

What also occurred to me, is that this process could be used, perhaps in a more refined version, to 'replace', or heck, generate dialog, having processed an actor's 'voice' from past movies, or other recordings...

http://gawker.com/5483834/roger-eberts-computer+generated-voice-is-far-better-than-we-imagined
 
I thought it sounded pretty "computerized" but?... In theory you could but why? You couldn't use anybody from whom you didn't get clearance. So having Bogart in your film isn't going to happen. You'd save time and a LOT of money hiring an impersonator instead.
Just an opinion.
 
Well, an actor is a lot more than a voice. Months of work go into building a role and no computer can simulate that.
 
Oh to answer the subject question. Not in the least. ADR is far more work than getting god sound in the first place.
 
ADR does not stand for "Automatic" Dialog Replacement. It stands for AUTOMATED Dialog Replacement.

Reminds me of one of our favorite office quotes: "If YOU don't have to do it, then it is *automated*."
 
Noiz2 said:
I thought it sounded pretty "computerized" but?... In theory you could but why? You couldn't use anybody from whom you didn't get clearance. So having Bogart in your film isn't going to happen. You'd save time and a LOT of money hiring an impersonator instead.
Just an opinion.
Click to expand...


What I was thinking of is along the lines of the current trends in motion capture, facial expression capture, that is using live actors, to then create better CGI characters.

Also, in the back of my mind, are the effects used in "Zelig"(1983), and also used in "Forrest Gump"(1994), where original footage is augmented to include current figures.

Alternatively, I was also thinking about situations where an actor's 'face' has been computer composited onto the stunt person's face, to get more 'realism' in the stunt.

Don't know how that would figure into voice work, other than perhaps an 'aging' actor/actress, voice crackling, and using previously recorded material to 'clean up' the voice.

At the moment, the Ebert example is 'too computerish', but I was interested in the process of dissecting a person's voice, and reconstituting it... and if there was anything going on in the audio world equivalent to the CG of the imaging world.
 
Richard Crowley said:
ADR does not stand for "Automatic" Dialog Replacement. It stands for AUTOMATED Dialog Replacement.

Reminds me of one of our favorite office quotes: "If YOU don't have to do it, then it is *automated*."
Click to expand...

Well, even with 'automated', it seems far more work than what I'd call 'automated'... the Boss is always asking me for the 'one button' solition... I keep saying, 'not in our life time'... at least with using computers... 'you're going to have to accept three buttons, and some text input'...

I've seen references use 'additional dialog replacement' which makes more sense to me as to the process and how it is used...
 
The A in ADR is actually not clearly defined a little like MOS has a few variations, and good stories to go with them. The Automatic, Automated etc, refered to the system of having film loops so the actor could keep doing the same scene with out having to wait for the projector to be rewound, and to the process of recording in sync to picture. You have to remember that this process came into being and got its name before magnetic recording. It was a LOT of work and not adjusting sync after the fact so you didn't need to go to far to with automation to make a big impact.

The only way that I can see the use of peoples voices to generate new lines (in a practical way) is when they are able to construct a virtual vocal path based on prior recordings, but even then actors are mostly "in character" when they are being recorded so how useful is it really. I mean your not going to get a good simulation of Tom Hanks voice if your taking samples from Forest Gump. I could see it for putting words in politicians mouths (in films) Having Roosevelt give a speech about aliens or what not, since those are the real voices and the settings are similar so the match would be closer.
 
You must log in or register to reply here.
Back
Top