KeyoAPI
CosyVoice3 API — Multilingual Zero-Shot TTS
CosyVoice3 is KeyoAPI’s async CosyVoice 3 TTS: multilingual zero-shot speech without cloning a repo or renting a GPU.
Overview
CosyVoice3 is KeyoAPI’s async CosyVoice 3 TTS: multilingual zero-shot speech without cloning a repo or renting a GPU. Call it with the same OpenAI-style key you use for Whisper and chat models. Async generation fits podcasts, course narration, IVR, and avatar pipelines that already wait on renderers. Prefer IndexTTS-2 when you need audio back inside a single HTTP round trip. About $1.37 per 10,000 characters. Map CMS script length to spend; confirm live rates on /pricing/CosyVoice3. Product voiceovers, localized support replies, game NPCs, and digital humans (pair with Duix-Avatar or InfiniteTalk). Customer bots that answer in the user’s language. POST /v1/async/audio/speech with model=CosyVoice3, then poll /v1/task/{id}. Docs: /brand/keyo-docs.html. If you want hosted inference instead of a research install, KeyoAPI is that path. Create speech jobs, poll until ready, then CDN-cache the audio URL. Offer clone samples only after consent. Fallback to a default voice when clone confidence is low. Measure time-to-first-byte of finished audio, not just API accept latency. Normalize text before synthesis. Store voice ids per locale. Fail closed on celebrity likeness without rights. Try it from /pricing/CosyVoice3. There is no separate upload playground.
Quick start
Base URL: https://www.keyoapi.xyz/v1
Endpoint: POST /v1/async/audio/speech
Model: model=CosyVoice3
curl https://www.keyoapi.xyz/v1/async/audio/speech \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"CosyVoice3","input":"Hello from KeyoAPI TTS."}'
# then: GET /v1/task/{id}Auth uses the same Bearer API key as chat. Full notes: Keyo docs.
FAQ
Is CosyVoice3 synchronous?
No. On KeyoAPI it uses POST /v1/async/audio/speech — create a task, then poll GET /v1/task/{id}. For a sync round trip, use IndexTTS-2 on /v1/audio/speech.
What is CosyVoice3 good for?
Multilingual zero-shot TTS with streaming-friendly async jobs, dialect and emotion/speed controls where the API schema exposes them.
How do I try it?
Open /pricing/CosyVoice3 after /sign-up. No separate upload playground — try from interactive pricing.
Price?
About $1.37 per 10,000 characters — confirm on /pricing/CosyVoice3.