Settings: Voice (TTS and speech recognition)
Text-to-speech providers, speech recognition providers, playback speed and reading options.
Settings → Voice has two lists. Add, edit, delete and select one provider in each. Test TTS speaks "Hello, this is a text-to-speech test."
Text to Speech#
| Provider | Key fields |
|---|---|
| System TTS | Uses the phone's TTS engine (default, no key). |
| OpenAI TTS | API key, base URL, model (e.g. gpt-4o-mini-tts), voice. |
| Gemini TTS | API key, model (gemini-2.5-flash-preview-tts), voice (e.g. Kore). |
| MiniMax TTS | API key, model, voice id, speed. |
| Qwen TTS | API key, model, voice, audio format, sample rate. |
| Groq TTS | API key, model (orpheus), voice. |
| xAI TTS | API key, voice, language. |
| MiMo TTS | API key, model, voice. |
| ElevenLabs TTS | API key, model, voice, stability, similarity boost. |
| Fish Audio TTS | API key, model, voice id, temperature, speed. |
| Step TTS | API key, model, voice, format, speed, sample rate, instruction. |
| xKiro Speech | Built-in key available, or use your own xKiro key. |
Speech Recognition (ASR)#
| Provider | Notes |
|---|---|
| Google (on-device) | Uses Google speech recognition built into your phone. No key. Optional language code like en-IN or hi-IN. |
| AssemblyAI | Ready out of the box with a built-in key; optional own key and language. |
| OpenAI Realtime | API key, WebSocket URL, model (gpt-4o-transcribe), language, prompt, VAD threshold, prefix padding, silence duration. |
| DashScope | Qwen realtime ASR. |
| Volcengine | Streaming ASR with resource id. |
| MiMo | Xiaomi MiMo ASR, sample rate, segment duration. |
| Step | StepFun ASR, ITN, timestamps, hotwords. |
| Groq | Whisper large v3 / turbo, language hint, sample rate, segment duration. |
Reading options#
In Chat and display → Chat → TTS Settings: Default playback speed (applied locally to every TTS provider), TTS Read Quoted Content Only, TTS Skip Bracketed Content, Auto-play TTS After Generation.