UT99.org | Unreal Tournament Community

Just thinking out loud. There are various services, paid for and open source that can generate "new" voice output when trained on a set of audio inputs.

So I wonder whether it would be possible to create a sound file containing all of the existing Announcer sounds, feed this to a training model, and then create a whole new set of authentic-sounding Announcer output saying - well - anything a UT content producer might want it to?

i think set too small for train good audio model. Can be wrong.

Adaptive generation is what I’d suggest.



AI shows some real impressive curation!!