Configuring Voice Providers

#047 0:20 Part of the Rysh video series

Deepgram or Whisper, sox or ffmpeg -- pick your transcription provider and recorder.

Configuring Voice Providers — 0:20 walkthrough

What you'll see

Show [voice_control] with tts_provider_name = "deepgram" (or "whisper") and api_key.
Show [voice] recorder = "auto" (sox/ffmpeg/afrecord/arecord) and max_seconds.
Show env-var overrides like RYSH_VOICE_CONTROL_API_KEY and RYSH_VOICE_RECORDER.
Note the macOS Microphone permission requirement on first use.

Commands shown

rysh
echo '[voice_control]'
echo 'tts_provider_name = deepgram   # or whisper (alias: openai)'
echo 'api_key = ...'
echo '[voice]'
echo 'recorder    = auto   # sox | ffmpeg | afrecord | arecord'
echo 'max_seconds = 120'
echo 'export RYSH_VOICE_ENABLED=true RYSH_VOICE_CONTROL_API_KEY=...'
echo '# macOS: System Settings -> Privacy & Security -> Microphone'

Keys used

Enter

Transcript

0:00Deepgram or Whisper, sox or ffmpeg -- pick your transcription provider and recorder.

0:03The voice_control section chooses your speech-to-text provider. Set tts_provider_name to deepgram -- the default -- or to whisper, and supply that provider's api_key.

0:13Back in the voice section, recorder picks how audio is captured. auto chooses the first tool on your PATH -- sox, ffmpeg, afrecord, or arecord -- and max_seconds caps a single recording.

0:22Prefer environment variables? Every key has one -- like RYSH_VOICE_CONTROL_API_KEY and RYSH_VOICE_RECORDER. On macOS, grant your terminal Microphone access on first use.

0:31Set the provider, set the recorder, grant the mic -- and your voice becomes just another way to prompt.

Key takeaway

Choose the transcription provider and audio recorder via config or env vars, then grant mic access to start dictating.