Talk mode

Talk mode is a continuous voice conversation loop:

1. Listen for speech
2. Send transcript to the model (main session, chat.send)
3. Wait for the response
4. Speak it via the configured Talk provider (
   text

   Copy

   talk.speak
   )

Behavior (macOS)

• Always-on overlay while Talk mode is enabled.
• Listening → Thinking → Speaking phase transitions.
• On a short pause (silence window), the current transcript is sent.
• Replies are written to WebChat (same as typing).
• Interrupt on speech (default on): if the user starts talking while the assistant is speaking, we stop playback and note the interruption timestamp for the next prompt.

Voice directives in replies

The assistant may prefix its reply with a single JSON line to control voice:

json
Copy
{ "voice": "<voice-id>", "once": true }

Rules:

• First non-empty line only.
• Unknown keys are ignored.
• text

  Copy

  once: true
  applies to the current reply only.
• Without
  text

  Copy

  once
  , the voice becomes the new default for Talk mode.
• The JSON line is stripped before TTS playback.

Supported keys:

• text

  Copy

  voice
  /
  text

  Copy

  voice_id
  /
  text

  Copy

  voiceId
• text

  Copy

  model
  /
  text

  Copy

  model_id
  /
  text

  Copy

  modelId
• text

  Copy

  speed
  ,
  text

  Copy

  rate
  (WPM),
  text

  Copy

  stability
  ,
  text

  Copy

  similarity
  ,
  text

  Copy

  style
  ,
  text

  Copy

  speakerBoost
• text

  Copy

  seed
  ,
  text

  Copy

  normalize
  ,
  text

  Copy

  lang
  ,
  text

  Copy

  output_format
  ,
  text

  Copy

  latency_tier
• text

  Copy

  once

Config (

text

Copy

~/.openclaw/openclaw.json

)

json5
Copy
{
  talk: {
    provider: "elevenlabs",
    providers: {
      elevenlabs: {
        voiceId: "elevenlabs_voice_id",
        modelId: "eleven_v3",
        outputFormat: "mp3_44100_128",
        apiKey: "elevenlabs_api_key",
      },
      mlx: {
        modelId: "mlx-community/Soprano-80M-bf16",
      },
      system: {},
    },
    speechLocale: "ru-RU",
    silenceTimeoutMs: 1500,
    interruptOnSpeech: true,
  },
}

Defaults:

• text

  Copy

  interruptOnSpeech
  : true
• text

  Copy

  silenceTimeoutMs
  : when unset, Talk keeps the platform default pause window before sending the transcript (
  text

  Copy

  700 ms on macOS and Android, 900 ms on iOS
  )
• text

  Copy

  provider
  : selects the active Talk provider. Use
  text

  Copy

  elevenlabs
  ,
  text

  Copy

  mlx
  , or
  text

  Copy

  system
  for the macOS-local playback paths.
• text

  Copy

  providers.<provider>.voiceId
  : falls back to
  text

  Copy

  ELEVENLABS_VOICE_ID
  /
  text

  Copy

  SAG_VOICE_ID
  for ElevenLabs (or first ElevenLabs voice when API key is available).
• text

  Copy

  providers.elevenlabs.modelId
  : defaults to
  text

  Copy

  eleven_v3
  when unset.
• text

  Copy

  providers.mlx.modelId
  : defaults to
  text

  Copy

  mlx-community/Soprano-80M-bf16
  when unset.
• text

  Copy

  providers.elevenlabs.apiKey
  : falls back to
  text

  Copy

  ELEVENLABS_API_KEY
  (or gateway shell profile if available).
• text

  Copy

  speechLocale
  : optional BCP 47 locale id for on-device Talk speech recognition on iOS/macOS. Leave unset to use the device default.
• text

  Copy

  outputFormat
  : defaults to
  text

  Copy

  pcm_44100
  on macOS/iOS and
  text

  Copy

  pcm_24000
  on Android (set
  text

  Copy

  mp3_*
  to force MP3 streaming)

macOS UI

• Menu bar toggle: Talk
• Config tab: Talk Mode group (voice id + interrupt toggle)
• Overlay:
  • Listening: cloud pulses with mic level
  • Thinking: sinking animation
  • Speaking: radiating rings
  • Click cloud: stop speaking
  • Click X: exit Talk mode

Android UI

• Voice tab toggle: Talk
• Manual Mic and Talk are mutually exclusive runtime capture modes.
• Manual Mic stops when the app leaves the foreground or the user leaves the Voice tab.
• Talk Mode keeps running until toggled off or the Android node disconnects, and uses Android's microphone foreground-service type while active.

Notes

• Requires Speech + Microphone permissions.
• Uses
  text

  Copy

  chat.send
  against session key
  text

  Copy

  main
  .
• The gateway resolves Talk playback through
  text

  Copy

  talk.speak
  using the active Talk provider. Android falls back to local system TTS only when that RPC is unavailable.
• macOS local MLX playback uses the bundled
  text

  Copy

  openclaw-mlx-tts
  helper when present, or an executable on
  text

  Copy

  PATH
  . Set
  text

  Copy

  OPENCLAW_MLX_TTS_BIN
  to point at a custom helper binary during development.
• text

  Copy

  stability
  for
  text

  Copy

  eleven_v3
  is validated to
  text

  Copy

  0.0
  ,
  text

  Copy

  0.5
  , or
  text

  Copy

  1.0
  ; other models accept
  text

  Copy

  0..1
  .
• text

  Copy

  latency_tier
  is validated to
  text

  Copy

  0..4
  when set.
• Android supports
  text

  Copy

  pcm_16000
  ,
  text

  Copy

  pcm_22050
  ,
  text

  Copy

  pcm_24000
  , and
  text

  Copy

  pcm_44100
  output formats for low-latency AudioTrack streaming.

Related

• Voice wake
• Audio and voice notes
• Media understanding