Replicate

Use Replicate's hosted TTS models as your provider.

Setup

In Settings → Admin → Providers:

1. Add a provider with type Replicate.
2. Enter your API key.
3. Set default model to: google/gemini-3.1-flash-tts (or your preferred model).

Users choose the provider, model and voice in Settings → Model & Voice.

Notes

• Built-in Replicate models:
  • google/gemini-3.1-flash-tts
  • minimax/speech-2.8-turbo
  • qwen/qwen3-tts
  • inworld/realtime-tts-1.5-mini
• Existing inworld/tts-1.5-mini settings run through the renamed model. Saved settings and audio cache identities are preserved; no migration is required.
• The former hosted Kokoro option is retired because of slow cold starts. Existing selections require another model for new generation; saved audio and settings are retained. New Replicate setups default to Gemini 3.1 Flash TTS with Zephyr. Local Kokoro remains available; Google AI Studio offers direct Gemini speech.
• You can also choose Other and enter any Replicate model ID (for example owner/model-name or owner/model-name:version).
• Native model speed is not available on all Replicate models; OpenReader hides/disables native speed controls where unsupported.
• Speech is requested by the compute worker, not the browser. The API key is never sent to browsers.

References

• Replicate
• TTS Providers
• TTS Environment Variables