// configure · 04

Voice (STT / TTS)

With the voice extra, the agent listens and speaks. Audio received on a channel is transcribed by local Whisper and becomes text for the harness; the reply is synth…

Open in full documentation GitHub

Enable

uv sync --extra voice    # faster-whisper (STT) + edge-tts (TTS)

Configuration

okami.yaml

voice:
  stt: { backend: whisper, model: small }     # local faster-whisper
  tts: { enabled: true, backend: edge, voice: pt-BR-AntonioNeural }

params voice.stt.backend enum whisper (local faster-whisper). voice.stt.model string Whisper model size (tiny…large). small voice.tts.backend enum edge (free) | minimax (token plan). voice.tts.voice string TTS voice (e.g. pt-BR-AntonioNeural).

CLI

okami transcribe <audio>transcribe an audio file (STT)

okami say "text"synthesize speech from text (TTS)

okami voiceturn-based voice conversation: microphone → agent → spoken reply

Channels & gatewayWhere voice is used.ConfigurationThe voice block in okami.yaml.

Open this page in the interactive docs