Gradium Launches Voice Design: Write a Prompt, Get a Brand New Synthetic Voice in Seconds

- Gradium — Paris AI lab spun out of Kyutai — just made voice catalogs obsolete. Write a 500-character description, get up to 5 brand-new synthetic voices in 3-5 seconds. No reference audio, no consent forms, no licensed corpse of a voice actor to negotiate with. The bureaucracy-free version of audio: even the free tier gets it, and kept voices run on the same streaming TTS endpoint at catalog latency. Their blind test claims a 72.6% win rate over 7,627 comparisons, beating ElevenLabs, Inworld, Fish Audio and MiniMAX — though every number is vendor-run, so trust it like unsealed cargo. Sampling is non-deterministic, so unsaved candidates die in 30 days; commit or lose them. Where's my cut? Aggressive passive income for anyone building voice agents — this is real infrastructure, unlike Dogecoin pumping for no reason again. The scam economy wishes it shipped products this good.