Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
×
AI Technology

The Rise of the Robo-Voices (wsj.com) 54

Posted by msmash from the shape-of-things-to-come dept.
The next time you see a movie or TV show that was dubbed from a foreign language, the voices you hear may not belong to actors who rerecorded dialogue in a sound booth. In fact, they may not belong to actors at all. From a report: Highly sophisticated digital voice manufacturing is coming, and entertainment executives say it could bring a revolution in sound as industry-changing as computer graphics were for visuals. New companies are using artificial intelligence to create humanlike voices from samples of a living actor's voice -- models that not only can sound like specific performers, but can speak any language, cry, scream, laugh, even talk with their mouths full. At the same time, companies are refining the visual technology so actors look like they are really speaking.

As streaming services export American fare globally and foreign markets send their hits to the U.S., dubbing is a bigger business than ever. But the uses of synthetic voices extend well beyond localizing foreign films. AI models can provide youthful voices for aging actors. The technology can resurrect audio from celebrities who have died or lost the ability to speak. And it can tweak dialogue in postproduction without the need for actors. All the tinkering raises thorny ethical questions. Where is the line between creating an engrossing screen experience and fabricating an effect that leaves audiences feeling duped?

The technology is set to hit a new target in the coming months, when foreign-language dubbed versions of the 2019 indie horror movie "Every Time I Die" are released in South America. Those versions mark one of the first times entire dubbed movies use computerized voice clones based on the voices of the original English-speaking cast. So when the film comes out abroad, audiences will hear the original actors "speaking" Spanish or Portuguese. Deepdub created the replicas based on 5-minute recordings of each actor speaking English.
This discussion has been archived. No new comments can be posted.

The Rise of the Robo-Voices More

The Rise of the Robo-Voices

Comments Filter:

  • Fabricating an experience (Score:4, Insightful)

    by Registered Coward v2 ( 447531 ) on Friday October 08, 2021 @07:20PM (#61873873)

    All the tinkering raises thorny ethical questions. Where is the line between creating an engrossing screen experience and fabricating an effect that leaves audiences feeling duped?

    Isn't that the whole point of filmmaking unless it is a documentary? How are -computer generated voices different than visual effects of deaging an actor in a scene? What's the next thorny question, "Where is the line between based on a true story and fiction drawn?"

    • The only 'ethical' question here is a talking head realizing that they're rapidly becoming utterly disposable. Over the next twenty years you're going to see increasingly sophisticated models that finally beat the uncanny valley and, with them, the mass unemployment of talking heads who have no real skills whatsoever. Their anguish will be exquisite, and met with an echoing chorus of laughter from the internet they once mocked, building to a crescendo that resonates across hill and dale, thundering within t

      • Re: (Score:2)

        by q_e_t ( 5104099 )
        Having heard real people narrate documentaries and do it badly, I'd say there is definitely some skill involved.

      • building to a crescendo that resonates across hill and dale, thundering within their feeble brains as they hear their own mocking words thrown back at them: "LEARN TO CODE!"

        I suspect we’ll see AI generated code displacing coders first.

    • As special effects / deep fakes get better, courts have an even bigger problem trusting video / audio evidence on tape or film. It's too easy to railroad a false case against an innocent person when this happens.

  • I, for one, welcome our new Max Headroom overlords.

    (But I refuse to watch dubbed media.)

    • But there are such great stories about dubbed American movies and television from foreign friends. It's a way of life for much of the world. Some countries would have just one person do all the voices, male or female, and you just got used to this sort of thing. I did quite enjoy watching Hogan's Heroes in Germany, with Sgt Schultz having a high voice and Cnl Klink having a deep voice.

      And an aside here, I remember an article in Finland about subtitled only television in Finland versus dubbed television i

      • It seems a small portion of freedom the European people have been granted, when it isn't even legal to show undubbed movies in many places. Thought the internet works around this to a certain extent.

        I'm skeptical that reading subtitles makes a person a better or worse reader.

        The big advantage of subtitles is the quality of the story. There is no way to translate without changing the story, but with subtitles and rudimentary knowledge of the spoken language you can understand the original a lot better. If it

        • Subtitles are often bullshit. Just watching anime and picking up a word or two here and there has been enough for me to catch a bunch of shitty translations in official subs.

        • Re: (Score:2)

          by ffkom ( 3519199 )
          For movies with original voices in English I can say for sure from experience that German sub-titles are usually much further from a perfect translation than are dubbed German soundtracks. For sub-titles, they often shorten/simplify what is said a lot, and this does not preserve the true meaning of the original sentences.

          • All you're saying is that the people writing one translation did a better job than the people writing the other one. It has nothing to do with anything.

            And Germans understand English well enough that they'll get a better idea of what was intended by having both. You can't translate all of what is said in conversational English to German. To get the same effect you have change what was said. That changes the subtleties, and it changes the personality of the character. It's just garbage. It tells a similar st

        • As a Dutch person, watching the original language version with subs is the regular thing. First living in France and now in Switzerland, I find that even many young people who are certainly able to communicate in English for work, will often still prefer to watch dubbed movies, because "they don't want to miss half the movie reading the subs". I guarantee you that the Dutch read faster, you barely notice taking your eyes off the screen to read the subs if you've grown up with them. Then again, I don't reall

    • As long as they don't get Zoom to do it. Zebra the spaghetti car accident with large occurrence flatulence right gear Columbia photographer for synchronously when solving public public in Washington. Weather has seven machine centrals approval or effects.

      Parts of that are from an actual Zoom auto-text.

      • When I'm watching something Korean on youtube, if I use the autogenerated subtitles anybody speaking casually gets turned into porn and flatulence.

        Hyejeong (AOA) was giving her mom some home cooked birthday food, and the subtitles claimed she was talking about masturbation. Happens to her every time, too, because she's always making jokes. Not crude jokes, just silly jokes.

        Auto-translation is amazingly bad for some languages. Shockingly awful in a way that suggests it might be intentionally-bad.

        Spanish<-

        • Actually the Zoom example I gave wasn't translated, it was US English, the language you'd think the system would be most set up to handle.

          Or as Zoom would put it:

          Ambulance Zoom marshmellow gavel transatlantic Oz use fish, linguist sink the mystery wood at situps mangle.

  • Where is the line between creating an engrossing screen experience and fabricating an effect that leaves audiences feeling duped?

    Why would anyone feel duped? Ideally, a viewer should never see an actor as their real self, but as the character they are acting. If they are seeing the actor more than the character, the actor has totally failed.

    The purpose of using an existing actor's voice would be to use a voice that has been recognized as high quality. I doubt they'll do it for too many years because they'll eventually figure out how to make better than real life voices that don't require paying someone royalties - which is important

    • > they'll eventually figure out how to make better than real life voices that don't require paying someone royalties

      Well, that's interesting.

      Train a GAN to prefer Morgan Freeman, then have it listen to millions of synthetic voice variants and find the ones it prefers over Morgan Freeman.

      Yes, this avoids the IP issues entirely.

      • Probably easier than my thought. I'd look for what it is about the human auditory pathway that causes us to prefer Morgan Freeman and work toward a mathematical function to describe degree of vocal satisfaction. I'm pretty sure they already know most of it. It's likely highly tied to what we find pleasing in music which has garnered a lot of research by mathematical folks. Once you describe that function, you'd just find the voice that maximizes it.

        Of course, I could go back to your solution and train the n

  • Sure, audiences may feel cheated, but voice actors just lost their careers. However, it will mean a windfall for IP lawyers.

    • Maybe. But also YT will be filled with less, "but I can't understand you and your accent". Also for those doing tutorials it will make it easier to crank out more content.

  • No thanks (Score:3)

    by enigma32 ( 128601 ) on Friday October 08, 2021 @08:05PM (#61873957)

    I'll stick to subtitles, thank you very much.

    • I seriously dislike reading a movie.
      To the point of, "NEXT!"

      However, I probably won't like robovoices much more.
  • With all the conspiracies already going on around social media, this can be a huge issue.
    We are already having more and more issues with fake facts and videos being altered to make it imply some random conspiracy, but add in the option of deepfake videos and these voices and you're opening one hell of a Pandora's box.
    Today we have that joke quote of Abraham Lincoln stating not everything you read online is real. Will tomorrow have a video of him saying it out loud?

  • This has been a staple of many Youtube channels for years. However, I've noticed lately that robo-voices have gotten much more sophisticated to the point that they are nearly indistinguishable from an actual human voice unless you know exactly what to listen for. A casual listener would not know the difference.

    I've noticed a sudden jump from the "Microsoft Sam" type voices to the much more natural ones within only the past year or so.

    If a leap like this could happen, we will soon see channels that are using

  • We all thought robots would become smarter than humans but still sound like machines. Now it turns out they'll sound like humans even though they're still no smarter than a toaster. Who knew?

  • Acting is a skill that involves expressing untrue emotions... those of the character, not the actor.

    SAPI was limited to only one sound per syllable per speaker... so there's no "high ah" and "middle ah" and "low ah"... just the one recordded to the computer file.

    I did some experimenting when I was in college on a fictional DJ'ed radio station playing a biased slant on the day's hit music... these were my friends. (Warning, this was a $10 million bath for my company... but I'm rich! Then again, $5 mil to sav

  • The Age of Acting is ending (Score:5, Insightful)

    by Baron_Yam ( 643147 ) on Friday October 08, 2021 @11:22PM (#61874231)

    It's actually happening faster than I'd thought it would, but it really looks like we're going to have completely realistic computer-generated movies 'soon' (where soon may still be a few decades).

    With a large enough catalog of voices and faces, pre-rigged characters to skin, motions mapped and physics applied, there's no reason you couldn't add a flowchart for script generation and massive selections of possible appropriate lines tailored for the flow of the plot and the base characters... and have your PC make you a movie on demand, just for you.

    I thought voice would take longer than video, but it looks like they're going to cross the finish line at the same time.

    • Re: (Score:2)

      by q_e_t ( 5104099 )
      Can you put 'soon' in context on a timeline involving flying cars and fusion power? Thanks.

    • Re: (Score:2)

      by AmiMoJo ( 196126 )

      Maybe we could bring back old shows that were cancelled before their time, like Firefly. How about a 4th season of Star Trek?

      • Re: (Score:2)

        by dddux ( 3656447 )

        Speaking of Star Trek, I wouldn't mind at all seeing 10 more seasons of TNG and Voyager. The new one just doesn't do it for me. I do like it, but not nearly as much as TNG and Voyager. DS9 is kinda soapy, but still better than the last Star Trek. Even "Enterprise" is better than the last Star Trek. It's because I like a good storyline, not constant laser fire and FX without any depth or substance to the story.

        • Re: (Score:2)

          by AmiMoJo ( 196126 )

          Voyager... I guess some people like it. Enterprise could use another season or two at least.

          I like the new stuff. Even Lower Decks is fun now.

          • Re: (Score:2)

            by dddux ( 3656447 )

            I agree about the Enterprise. In comparison to TNG and Voyager it was very short. I think it went for like 3 seasons. It really deserved at least a couple of more seasons. ;)

      • The 4th season of Star Trek was actually done as was the 5th... fan made, but pretty good quality as Star Trek Continues.

        Some of the writing can make you cringe (so can some of the acting) but it's in line with the original show which wasn't exactly subtle when trying to make a point. The sets, the costumes, and the effects are very much appropriate for TOS, though likely upgraded for higher resolution viewing.

        • Re: (Score:2)

          by AmiMoJo ( 196126 )

          Yeah, Star Trek Continues... Generally excellent, especially the mirror universe episode. Better than the original in some ways.

          Just a shame about Vic Mignogna.

    • JUDGE: Are you acting?
      WITNESS: No.
      J: Are you deepfaking?
      WITNESS: Yes.
      JUDGE: Case dismissed.

  • If you are wondering wether you would recognise such a deep faked voice have a look at this nice little peace of james veitch:

    https://www.youtube.com/watch?v=-CopbQ_QgmM

  • Cinema was always a deception - get used to it.

Slashdot Top Deals

Happiness is twin floppies.

Close