Superhuman Speech
·
Zach Ocean
Sometimes it feels like superhuman text to speech (and generative audio in general) is solved.
But it’s not—even though TTS models are close or even superior to average humans in tasks like audiobook narration, they’re still way worse than the best humans in most speech domains.
What follows is a collection of links that I believe represent the highest level of human speech generation.
This list was made off the top of my head in just a few minutes, so please send me more that come to mind.
Also, please send me examples of AI-generated outputs or demos that approach or exceed these in terms of quality. Contact me here.
Monologue
Coffee Is For Closers
Drill Sergeant - Full Metal Jacket
Patrice O’Neal Breaks Down Radiohead’s Creep
Eric Thomas (the Hip Hop Preacher)
Raw
Produced