Adobe’s wild ‘Photoshop for audio’ experiment can change what you said

“Forget about making a shark look like it’s attacking a diver or faking a David Hasselhoff photobomb,” Ian Paul writes for PCWorld. “In Adobe’s future you’ll be putting words in people’s mouths.”

“During the Adobe Max conference the company showed off eleven experimental projects, dubbed ‘Sneaks,'” Paul writes. “The showcase included a new audio editing feature called Project VoCo, which allows for ‘Photoshopping voiceovers.'”

“VoCo allows you to take a piece of audio and manipulate it in a number of ways just by editing text,” Paul writes. “You can rearrange words to change a sentence, or even insert words that weren’t said in the original recording.”

Read more in the full article here.

MacDailyNews Take: Political ad producers are salivating like St. Bernards.


  1. Having watched the whole video, two things may reassure those who are freaking out about the possibilities.

    First, much like photoshoping, it can be detected. Poorly done work can be detected bay anyone; well done work can be detected by experts, but there is no doubt, it will be detectable;

    Second, Adobe is purposely building in some watermarking and other features in order to make sure it is always detectable by those who need to tell the difference (presumably, law enforcement / justice system, legal experts, etc.).

    However, much like what Photoshop enabled, this too will make life a lot easier for many professionals working with spoken audio.

    1. As there are fine art copies that can’t be detected with certainty even with all the technology at their disposal, by life long experts let’s say I don’t share your certainty. As a long time Photoshop user with not a little skill to claim that photoshop fakes are always detectable is equally optimistic. Hell there are even ludicrous debates over clearly faked UFO pics let alone highly skilled Photoshop imagery. The thing is, even if you can detect fakes you have the endless problem of convincing others especially the masses that they should believe any such claims. Wish you luck on that one, in the world of rampant conspiracy theorists.

      1. You are absolutely right.

        My thinking wasn’t as much about the ultra-[left/right]-wing “news” sites (none of which produce actual, factual news), but more along the lines of using such recordings as evidence in the judicial process. The point being, a proper expert could dismiss such fake evidence, much in a way today’s photoshopped evidence is dismissed by experts.

        People will believe what people want to believe, and they didn’t need photoshopped images in order to believe it, so that won’t change.

      2. Agreed. Even if the “soundshop” work isn’t expert, FUD will spread like wildfire across the internet. And exposing the fraud will not repair the damage. People tend to believe what they want to believe and will latch on to any “evidence” supporting those beliefs.

    1. Look, I’m not voting for either. (Maybe Gary Johnson will get my protest vote.) But how can you be against Hillary for lying (which she is certainly guilty of) and ignore Trump’s bald faced fabrications? Pot, meet kettle!

  2. I can see this being HUGE in the ADR market (automated dialogue replacement) for film and video. Now, actors must routinely come in and record over their own dialogue to make up for room noise, glitches or script changes. With this technology, you could replace the audio without them. Or write new dialogue for actors who have died or who are unavailable.

    Perhaps in 10-15 years, you’ll just have actors come in and be 3D-scanned and have them record some dialogue, and let the computer complete their performance. Pretty sure there was a film that explored this, but can’t remember the name.

    1. Decades ago, and I mean way before computers were everywhere, someone wrote a SF short story about just this sort of thing. It appeared in some anthology. It dealt with it from the perspective of an actor and his fall from grace because he was no longer needed to actually perform.

      Can’t remember the title or author, unfortunately. Think it was written in the fifties. Very prescient.

  3. With how powerful iPhones have become and will become, it’s not hard to imagine a future when you could use a real-time voice filter that will recognize what you are saying and synthesize someone else’s voice. So you could make phone calls using someone else’s voice. It’s not hard to imagine this happening when you look at the real-time video filters.

Reader Feedback

This site uses Akismet to reduce spam. Learn how your comment data is processed.