Voice
Speak to your AI and hear it respond — TTS, speech-to-text, and hands-free voice sessions.
Speak to your AI and hear it respond. Twent supports configurable text-to-speech and speech-to-text engines, plus hands-free voice sessions.
Text-to-Speech Engines
| Engine | Type |
|---|---|
| System TTS | Built-in Android |
| OpenAI TTS | Cloud |
| SiliconFlow TTS | Cloud |
Configure under Settings → Speech Services.
Speech-to-Text Engines
| Engine | Offline |
|---|---|
| Sherpa-NCNN (on-device) | Yes |
| OpenAI STT | No |
| Deepgram STT | No |
Cloud STT engines automatically fall back to on-device recognition when a network connection isn’t available.
Hands-free voice
Start a voice session and speak your command — the AI listens, transcribes, responds with its own voice (“Wanna get something done?”), and can drive your screen while talking. The overlay agent (home/power long-press) uses the same voice pipeline.