TechSpot

Adobe is developing an audio app that literally lets you put words into someone's mouth

By Shawn Knight
Nov 6, 2016
Post New Reply
  1. Adobe at its recent MAX 2016 conference for creative professionals demonstrated an experimental technology that’s sure to stir up its fair share of controversy.

    Known internally as VoCo and currently in development with Princeton University, the technology in question can be best described as Photoshop for audio. As Adobe developer Zeyu Jin showcases in the clip above, you can rearrange the order of spoken words and literally put words in someone’s mouth to make it sound as if they said something that never actually happened.

    The current iteration of the technology needs roughly 20 minutes of dialog from a person in order to recreate their voice. Given the proliferation of audiobooks, podcasts, vlogs and so on, finding enough material to feed the program – especially if the target is a celebrity, public figure or social media influencer – would be trivial.

    It’s easy to imagine how technology of this nature could be used in all sorts of unethical or nefarious manners but according to Jin, the development team has researched how to prevent forgery (likening it to watermarking for images).

    As for legitimate uses, Adobe said in a companion blog post that when doing voiceover, dialogue and narration work, it’d be nice to have the option to edit or insert a few words without the hassle of recreating the recording environment or bringing the artist back in for a follow-up session.

    Adobe hasn’t yet said when or even if the technology will one day make its way into a consumer-facing product. If it does, however, we’ll have to condition ourselves to be skeptical of any audio we listen to (just as we do today with images thanks to Photoshop).

    Permalink to story.

     
  2. Uncle Al

    Uncle Al TS Evangelist Posts: 1,675   +779

    Interesting and yet another way to allow criminals to successfully fool those that are not tech savvy .....
     
    Reehahs and TheDreams like this.
  3. Kenrick

    Kenrick TS Booster Posts: 188   +88

    Better fix flash first before releasing another crime tool.
     
  4. Evernessince

    Evernessince TS Evangelist Posts: 1,194   +585

    This is one of those tools that has far more bad uses than good.
     
  5. cliffordcooley

    cliffordcooley TS Guardian Fighter Posts: 8,555   +2,898

    I'm sure we will all continue to believe everything we hear. This will change nothing because naturally we want to believe lies!
     
  6. fktech

    fktech TS Booster Posts: 118   +38

    This isn't new.
     
  7. mbrowne5061

    mbrowne5061 TS Evangelist Posts: 338   +133

    Cutting in word someone's already spoken? No.
    But this takes words they never said, generates how it would have sounded had they said it, and then cuts it in.
     
  8. Win7Dev

    Win7Dev TS Evangelist Posts: 567   +174

    This is pretty cool and I hope that VoCo becomes apart of Soundbooth/Audition
     
  9. Moneyd623

    Moneyd623 TS Member

    Cool, maybe now we'll be able to preserve the best story telling voices and use them for as long as we'd like. Would be cool if I could set it up to have someone like Morgan Freeman be the voice for all my audio books or something like that.
     
  10. rvnwlfdroid

    rvnwlfdroid TS Booster Posts: 135   +23

    I'd pay to be able to do that. I can't wait to hear a longer generated clip.
     
  11. Igrecman

    Igrecman TS Enthusiast Posts: 91   +47

    The watermark to be effective would need to be some kind of frequency the human ear can't hear all over the modified words. If it's something else, I guess it would be easy to bypass by re-recording the result with Audacity.
     

Similar Topics

Add New Comment

You need to be a member to leave a comment. Join thousands of tech enthusiasts and participate.
TechSpot Account You may also...