# Open WebUI text to speech for chat replies

Open WebUI users who read in more than one language need replies read aloud in each of them. Set Open WebUI's built-in OpenAI audio engine to Paxa with a base URL and an API key, and replies play aloud in English, Thai, and Mandarin. The same settings add dictation through Paxa transcription.

## The OpenAI engines, pointed at Paxa

A deployment sets both engines with environment variables, and the Admin Panel takes the same fields by hand. The field tables are the setup guide's own.

## From a key to replies read aloud

Open WebUI keeps its models and chats. Paxa reads replies aloud and transcribes the messages users dictate.

1. **Open the audio settings**: Go to Admin Panel > Settings > Audio, where Text-to-Speech and Speech-to-Text each take an engine.
2. **Point speech at Paxa**: Choose the OpenAI engine, set API Base URL to https://api.paxalabs.com/v1, paste your Paxa API key, and set TTS Model to paxa-tts-flash-v2 and TTS Voice to a Paxa voice id.
3. **Add dictation**: Under Speech-to-Text, choose OpenAI with the same base URL and key, set Request Format to Multipart Upload, and set STT Model to paxa-stt-lite-v1-preview.

Paxa TTS Flash v2 lists at $10 per 1M characters. Paxa STT Lite lists at $0.50 per hour of audio, billed by the millisecond.

## Questions

### What is Open WebUI?

Open WebUI describes itself as a self-hosted AI platform that is extensible, feature-rich, user-friendly, and built to run entirely offline. It works with Ollama and OpenAI-compatible APIs, and its audio settings accept OpenAI-compatible speech engines.

### Which voices can Open WebUI use?

Open WebUI can use any of the 50 Paxa TTS Flash v2 voices, and each one speaks English, Thai, and Mandarin. The TTS Voice suggestions list them from [GET /v1/audio/voices](https://paxalabs.com/docs/api/audio-voices), and [the voice library](https://paxalabs.com/voices) plays them.

### Why does a chat keep its old voice?

A voice set on a model, or in a user's own settings, takes priority over the admin default. Change those too, and leave Additional Parameters empty.

### How long can a dictated message run?

A recording can run up to 120 minutes. STT Model starts empty in Open WebUI, and a request without a model answers 400. Paxa STT Lite is a Research preview.

## Start with free credits

Sign in with Google, GitHub, or Hugging Face and spend 100 one-time free credits on your own text.