SillyTavern Setup
Configure Fish Audio OpenAI-compatible TTS in SillyTavern and assign different voices to characters and narration.
Connect Fish Audio TTS to SillyTavern
SillyTavern's TTS extension includes an OpenAI Compatible provider. It accepts a complete speech endpoint, model, available voice list, and a Voice Map for each character.
Open the TTS extension
Open SillyTavern's Extensions panel, select Text-to-Speech, enable TTS, and choose the OpenAI Compatible provider.
Enter:
| Setting | Value |
|---|---|
| Provider Endpoint | https://fishaudio.org/v1/audio/speech |
| API Key | Your Fish Audio API key |
| Model | fishaudio-s21pro-flash |
| Available Voices | 00a1b221-6137-4b73-ad62-b0cbce134167 |
| Default Voice | The same test Voice ID |
SillyTavern asks for the full Provider Endpoint, so include /audio/speech here. This differs from apps that request only a Base URL.
Map a character voice
In Voice Map, select the current character, map it to the test voice, and click Apply. Generate a short character reply and use message playback or automatic TTS to test it.
After the connection works, add multiple Fish Audio Voice IDs to Available Voices and assign them to characters, narration, or quoted text. Find IDs through the voice list API.
Troubleshooting
- User not in voice map: select a voice for the current character and click Apply.
- 404: use the complete
https://fishaudio.org/v1/audio/speechendpoint. - No voices are available: manually add Fish Audio Voice IDs to Available Voices.
- Speech plays twice: check whether automatic generation, message playback, and narration are enabled together.
- 401/402: verify the API key and separate API credit balance.