SillyTavern text to speech with a voice for each character
A SillyTavern chat with several characters needs a distinct voice for each one, in the language each one speaks. Connect SillyTavern's OpenAI Compatible TTS provider to Paxa with an endpoint and an API key. Each character then speaks English, Thai, and Mandarin in the Paxa TTS Flash v2 voice you assign.
One provider, a voice per character#
The connection steps and the feature table are the setup guide's own. SillyTavern reads its voice list from the Available Voices field, and the voice map assigns one to each character.
Connect
- Open the TTS settingsOpen the Extensions panel and its TTS section.
- Choose the providerSelect OpenAI Compatible as the TTS provider.
- Set the endpointSet Provider Endpoint to
https://api.paxalabs.com/v1/audio/speech. The field takes the full URL. - Add the keyClick API Key and paste your Paxa API key.
- Set the modelSet Model to
paxa-tts-flash-v2. - List the voicesSet Available Voices to Paxa voice ids separated by commas, such as
khanomkrok,nomyen,foithong. - Set the speedKeep Speed between 0.5 and 1.5.
- Assign voicesCheck Enabled, open a chat, choose a voice for each character in the voice map, and click Apply.
Feature support
| Feature | Support |
|---|---|
| Voice list from the API | No. SillyTavern reads the voices from Available Voices. |
| Speed | Yes, from 0.5 to 1.5 |
| Audio format | MP3, which SillyTavern always requests |
| SillyTavern's defaults | Yes. tts-1 and the default voices map to Paxa voices (accepted voice names). |
From a key to a cast with voices#
SillyTavern keeps its characters and its text model. Paxa reads each character's lines in the voice you assign.
Choose the provider
In the Extensions panel's TTS section, select OpenAI Compatible as the TTS provider.
Connect Paxa
Set Provider Endpoint to the full URL https://api.paxalabs.com/v1/audio/speech, paste your Paxa API key, and set Model to paxa-tts-flash-v2.
Assign voices
List Paxa voice ids in Available Voices, check Enabled, and choose a voice for each character in the voice map.
Paxa TTS Flash v2 lists at $10 per 1M characters.
Questions#
What is SillyTavern?
SillyTavern describes itself as a locally installed user interface for text generation LLMs, image generation engines, and TTS voice models. It sends each speech request from its own server, which needs to reach the Paxa API.
Which voices can characters use?
Characters can use any of the 50 Paxa TTS Flash v2 voices, and each one speaks English, Thai, and Mandarin. Pick ids from GET /v1/voices, or listen first in the voice library.
Do SillyTavern's default voices work?
Yes. tts-1 and SillyTavern's default voices map to Paxa voices (accepted voice names). List Paxa voice ids to choose the voices yourself.
Why does a high Speed setting fail?
SillyTavern's Speed slider reaches past Paxa's range. Paxa accepts 0.5 to 1.5 and answers 400 outside it.
Start with free credits
Sign in with Google, GitHub, or Hugging Face and spend 100 one-time free credits on your own text.
Products
Use cases