Straight Outta Comp-Sci: now AI voice actors can rap | PC Gamer - lundyablumersy

Uncurving Outta Comp-Sci: now AI voice actors can rap

(Image credit: Replica)

Along his track Rap God the effervescently witty Eminem takes objective at incomparable joint criticism of his delivery style: "They said I rap like a robot so birdcall me Rap-bot—Only for Pine Tree State to whan like a computer essential beryllium in my genes, I got a laptop computer in my back sac..." and so on.

It's a fairly common criticism in hip joint-hop on, applied to many less famous than Eminem. The idea of a 'robotic' delivery style is a big diss: your delivery is monotonous, the rhymes are obvious, you don't please the spike.

Welp, looks wish we may need to Ra-believe things, thanks to an AI voice company called Replica and the semi-cursed result of its employees messing around in what was evidently a 'hackathon' (thanks, Verge).

Remarkably enough, the blame itself comes from ace individual, stock split across the various Artificial intelligence actors. Replica employee Shreyas told PCG that "One Replica team member performed all the rap music bits. After passing his source recording done our arrangement, we then can apply the tone/cadence/performance onto any of our 50+ (and ontogeny) AI voices."

It's got some funny lines, too.

"And then maybe you can see that I'm a virtual actor,

And I'm break new ground like a literal tractor."

Non bad, though what would make that perfect would follow an AI with the voice of Alan Partridge, Norfolk's greatest Disk-jockey.

"The video itself is the result of something produced internally during a hackathon, victimization a feature that's currently nether development," explains Shreyas. "The feature itself is purely all but applying the same cadence/timing/delivery from a source transcription onto any of our AI voices. The fact that IT can actually be pushed indeed furthest as to pick off a rap is very cool—and an undesigned issue of the feature that we were originally only intending for adding perfectly timed lip-synchronize to a video or 3D animation. For instance, here's the aforesaid feature applied to Unreal Engine's Metahumans."

The feature is collect to establish later this twelvemonth, after debuting at GDC in July. The idea is basically to make water it possible for developers to have a large cast of vocalism-acted characters, with accurate lip-synching and intonation, without having to go through the process of individually invigorating and transcription every line. Depending happening the results it could follow a huge advance for small developers: we'ray all laughing at the realistic rappers in real time, but in a few years this technical school could be in half the games we play.