The frontier of voice synth has reached roughly human parity. It’s also very inexpensive.
I’d be shocked if we don’t have mainstream games with AI voices soon.
I’d be shocked if we don’t have mainstream games with AI voices soon.
I want to use voice synthesis to read my blog articles out loud to me as a self-editing step, but the experience was nothing like a human when I last tried.
The voices sounded natural-ish alright, but intonation was all over the place.
On https://artificialanalysis.ai/text-to-speech/models, you can plot speech preference ELO and pronunciation correctness against price.
Gemini 3.8 Flash / Flash-Lite TTS seem good, for example, they lie on the frontier of both.