Wednesday, November 16, 2016

New tools will let you edit and invent spoken words



Adobe recently gave a sneak peak of new software called #VoCo with an amazing (and potentially alarming) capacity to edit recorded speech.

Users will easily be able to change word order and even to invent new words using a simple keyboard interface. Novel words can be generated if you have a database of about 20 minutes of recorded speech to draw from.  (Link to YouTube).

A few implications:
1. Fake news stories with manufactured quotes will be easy to create.
2. Human voice actors won't be need to read an entire recorded book or to voice every line for an animated film, especially for low budget productions.
3. Adding voices in ADR to the sound edit of a film will be much easier.

10 comments:

Tom Hart said...

I wonder why, visual artist that I am, the implications of this bother me way more than developments about image copying and manipulation. Probably it's the potential application to manipulate newscasts.

Robert Michael Walsh said...

Do you trust me, or your lying eyes and ears?

Jennifer Branch said...

Just a bit disturbing.

A Colonel of Truth said...

For all the good comes bad, too. With the human brain remaining the filter, the discriminator. Stephen Hawking is in the news today predicting humans extinct in a 1,000 years. Not much time for everyone to get their act together. Buy gold!

René PleinAir said...

It does make me think about Propaganda - Dream Within A Dream or would it become a nightmare?

> https://www.youtube.com/watch?v=P6wzXzI-IOE <

Beth said...

The potential for character assassination is immense. Got a tech savvy enemy? What words couldn't they put in your mouth?

Gavin said...

One step closer to postmodernity and Baudrillard's ideas on Simulacra and Simulation. And at a time where faith in things like the mainstream media seem to be at an all time low. I can see positive applications for this, but the more we simulate reality as we know it, the further we seem to unravel. Perhaps why I'm an old stick in the mud who likes simple things like a canvas and brush or a pencil to paper (he says typing at a computer whilst connected to the Internet!).

Will Dudla said...

What is crazy about this is that it seems to be a step up from Vocoloid, which has been around for quite a while.

https://en.wikipedia.org/wiki/Vocaloid

You can google a ton of songs on Youtube that use the Vocoloid technology. There are also several live concerts of holographic characters called Hatsune Miku singing songs in Japan.

Wild and crazy times we live in!

I do find some solace in the "watermarking" comment that Mr. Jin mentions though (even though I am sure that could be hacked without a problem).

Tobias Gembalski said...

In conjunction with the progress in picture manipulaiton this can be quite frightening for an indivual artist. Maybe from now on one should tag the created art (drawing, video etc) with a seal of quality, like: "artistic content handcrafted by an individual human being".

Jayson Mondala said...

Just as manipulated digital photos can be forensically examined, there's probably an analog to audio. It would perhaps call into question voice recordings as evidence, or at least create a new lineup of expert witnesses for attorneys to hire. It could help a computer pass Turing Tests and Uncanny Valleys a little easier, though.