Voice Synthesis Technology

Question

Does anyone around here know what the state of the art software is in Hollywood these days as far as voice synthesis goes?

It may be a reality already, I don't know, but I've envisioned over the past few years, being able to type in a line of dialogue into a script I'm working on, and then cutting and pasting that into another program, or clicking a button, which converts the dialogue into sound, from a dropdown list of currently available pre-defined actor voices.

Being a computer programmer, I know there's gotta be a way to write a program that encodes a person's speech patterns into some kind of database, and then uses the database to replicate the person's voice.
It's complicated programming for sure, and I'm not good enough to do it myself, but I figured somebody out there's gotta be working on something like this.
This would be especially useful in cartoons, where right now living people have to read the lines. What if their voices and voice patterns could be encoded for future use, even after they die?

What if the guy who does the Homer Simpson voice gets hit by a bus tomorrow? Do they have a backup guy or stand-in who sounds just like him, ready to take his place? Maybe a digital backup of the guy's voice pattern is what's needed?
How close from sci-fi is this type of thing, anyone know?

Furthermore, what would the copyright complications be? If anyone who buys someone else's voice pattern database can imitate them, digitally, then all kinds of things might happen. Could people sell their voice patterns for perpetual income, to their families, after they die?
You have to admit, it'd be pretty cool to be typing some dialogue into a script and then hit a hotkey or press a button and HEAR the dialogue being spoken by some famous actor or cartoon character. The parenthetical could be used to tailor the inflections, like (softly), (loud),

, etc. Special editing software would allow you to edit the inflections on each word, if you needed to.

How close are we to this? A million miles or more?

Elroy Willis
www.elroysemporium.com

Usenet · Accepted Answer

[nq:1]Does anyone around here know what the state of the art software is in Hollywood these days as far as ... each word, if you needed to. How close are we to this? A million miles or more? Elroy Williswww.elroysemporium.com[/nq]
Given how far we've come in generating completely photo-realistic artificial images in CGI, one would have thought that creating or imitating voices would have been a breeze, and yet it's clear that the opposite seems to be the case.
Even generating a believable artificial voice from scratch certainly you've heard them has proven to be an extremely difficult task. Those that are even moderately convincing are actually generated from sampling real voices those that you find on those annoying voice answering systems that say things like, "Okay, I didn't quite hear what you said," to which you answer, "I said I wanted to track you down and *** erase you, you *** annoying recording put a human being on the line!"
The "real" thing that is, artificial voices that do things like scan text and read it back well, certainly you've heard such things and they're really terrible. Not even close to a believable human voice.
I suppose it might be possible ****, given enough money and motivation, why wouldn't it be? It's really all a matter of developing the proper tools to manipulate the variables that go into generating the human voice, just as we've developed the various tools that go into generating all the variables that go into producing photo- realistic CGI images.
I suppose the reality is, though, there are currently enough people who could, in a pinch, imitate Homer Simpson's voice that it probably wouldn't be worth it.
The market forces that propelled the CGI revolution really had to do with producing images that couldn't be successfully produced otherwise.
But if you really wanted to have a scene with Bogart's voice you could find somebody to do it almost certainly as convincingly as you could by generating it by way of computer.
Conversely, if you had an actor die during the making of your movie so that you needed somebody to finish dubbing in his lines it's really too late in the day to start a major research project in terms of generating the technology to "manufacture" his voice digitally. You're really be stuck trying to find somebody to imitate his voice (in fact, in David Niven's last movie, that's exactly what happened. He died and Rich Little had to come in and finished dubbing in the lines for the last few scenes that he couldn't complete).
NMS

Usenet · Answer

[nq:1]Given how far we've come in generating completely photo-realistic artificial images in CGI, one would have thought that creating or imitating voices would have been a breeze, and yet it's clear that the opposite seems to be the case.[/nq]
Voice recognition is still in its infancy as well. It's one of those things humans do really well, but is hard to reproduce with computers.
[nq:1]Even generating a believable artificial voice from scratch certainly you've heard them has proven to be an extremely ... well, certainly you've heard such things and they're really terrible. Not even close to a believable human voice.[/nq]
I can't believe Stephen Hawking doesn't have a better sounding computerized voice. Have you heard it? Sounds like the voice used for the computer in the War Games movie 20 years ago.
[nq:1]I suppose it might be possible ****, given enough money and motivation, why wouldn't it be? It's really all ... we've developed the various tools that go into generating all the variables that go into producing photo- realistic CGI images.[/nq]
I thought maybe it might take an intricate simulation of the vocal chords and lungs and lips and tongue and teeth moving to produce the sounds, rather than just trying to duplicate some speech patterns.

The other way would be to extract all the phonemes necessary from a living person's voice, and use them to reproduce words which are spelled phonetically in some data dictionary.
Either way, it's some very complicated programming.
[nq:1]I suppose the reality is, though, there are currently enough people who could, in a pinch, imitate Homer Simpson's voice that it probably wouldn't be worth it.[/nq]
Any idea how much people get paid for doing voice-overs for a cartoon series like the Simpsons?
[nq:1]The market forces that propelled the CGI revolution really had to do with producing images that couldn't be successfully produced ... could find somebody to do it almost certainly as convincingly as you could by generating it by way of computer.[/nq]
You're probably right.
[nq:1]Conversely, if you had an actor die during the making of your movie so that you needed somebody to finish ... Little had to come in and finished dubbing in the lines for the last few scenes that he couldn't complete).[/nq]
I didn't know they called in Rich Little. I haven't heard of him in years. Thought he might even be dead. I used to see him all the time on the Johnny Carson show when I was a kid.

Elroy Willis
www.elroysemporium.com

Usenet · Answer

[nq:1]Does anyone around here know what the state of the art software is in Hollywood these days as far as ... read the lines. What if their voices and voice patterns could be encoded for future use, even after they die?[/nq]
If you listen to the commentary track on animated movies and series, you'll occasionally hear that they do final animation AFTER the voice track is recorded, to take advantage of an actor's reading. For some reason, Eddie Murphy's character in Shrek comes to mind in this regard.

I don't doubt that they could do some cartoons with computer generated voices (the vocal track on Jonny Quest always sounded computer-generated to me, even as a kid), but I think we're stuck with real actors for now (thank ***).
As Mr. Stevens says, they're much more likely to get a substitute actor to replace the voice of an original who couldn't complete the movie. I hadn't heard the David Niven story, but the very distinctive voice of Jack Hawkins was dubbed after he lost his voice due to throat cancer.

So it would be fun to have realistic voices in Hollywood Screenwriter or Final Draft, but I don't think screenwriters constitute a big enough market.
[nq:1]What if the guy who does the Homer Simpson voice gets hit by a bus tomorrow? Do they have a ... the inflections on each word, if you needed to. How close are we to this? A million miles or more?[/nq]
Mark Cipra
"Know your lines and don't bump into the furniture." (Spencer Tracy on acting)
(Play Indiana Jones! Hide the "ark" in my address to reply by email)

Usenet · Answer

[nq:2]This would be especially useful in cartoons, where right now ... could be encoded for future use, even after they die?[/nq]
[nq:1]If you listen to the commentary track on animated movies and series, you'll occasionally hear that they do final animation ... take advantage of an actor's reading. For some reason, Eddie Murphy's character in Shrek comes to mind in this regard.[/nq]
They have to animate the actual lip and mouth and maybe even tongue positions to get the video to match the audio, right?
[nq:1]I don't doubt that they could do some cartoons with computer generated voices[/nq]
If the generated voice was introduced in the first episode, then nobody would know the difference. An actual human being wouldn't be needed after that..
[nq:1](the vocal track on Jonny Quest always sounded computer-generated to me, even as a kid), but I think we're stuck with real actors for now (thank ***).[/nq]
Johnny Quest, wow, that brings back some memories. What kind of dog was Bandit, anyway? A pug?
[nq:1]As Mr. Stevens says, they're much more likely to get a substitute actor to replace the voice of an original ... to have realistic voices in Hollywood Screenwriter or Final Draft, but I don't think screenwriters constitute a big enough market.[/nq]
Plus, while writing, writers aren't supposed to have specific actors in mind while they're writing their dialogue...

Elroy Willis
www.elroysemporium.com

Usenet · Answer

[nq:2]True, but as others have pointed out, the technology just isn't there yet.[/nq]
[nq:1]I think it will be, maybe, some day.[/nq]
It might work for cheap Saturday morning cartoons, but a generated voice will never have all the inflections of a real human voice - not unles you get that computer to feel and express human emotions. And getting a technician to direct the computer to express the emotion reqired by a line becomes more complicated than just paying the voice actor scale in a studio.
It's a lot like saying that computer-generated characters will one day replace real actor. Well, one simple observation of the star system reveals the pretty obvious fact that people want stars to idolize.
Again, as is the case with all predictions about future technology: it's not because we have the capability that people will necessary adopt it.
jaybee

Usenet · Answer

[nq:2]I think it will be, maybe, some day.[/nq]
[nq:1]It might work for cheap Saturday morning cartoons, but a generated voice will never have all the inflections of a ... actor. Well, one simple observation of the star system reveals the pretty obvious fact that people want stars to idolize.[/nq]
And peek into their *** lives, as well.
[nq:1]Again, as is the case with all predictions about future technology: it's not because we have the capability that people will necessary adopt it.[/nq]
It might work for some cartoons, but that's about it, I think. I personally don't even know what the guy who does the Homer Simpson voice looks like, nor do I really care. If somebody could realistically re-create his voice using a computer, it wouldn't really matter to me.

Elroy Willis
www.elroysemporium.com

Voice Synthesis Technology

Top answer

Get the Weekly English Kit 📬

Related Questions