Claude/Code/Q/Does Claude Code have a dictation or voice-to-text command like Claude Desktop?
- AI/Response
- Short answer: Yes — Claude Code has native voice dictation via the
/voice command, available since v2.1.69. - How to enable:
- Type
/voice in the Claude Code prompt to activate - Two modes:
- Hold mode (default): hold
Space while speaking, release to finish - Tap mode (v2.1.116+): tap
Space once to start, tap again to send — enable with /voice tap
- Key differences from Claude Desktop:
- Claude Desktop uses
⇪ (Caps Lock) as the dictation shortcut - Claude Code's default trigger key is
Space; rebind in ~/.claude/keybindings.json (e.g. to meta+k) - No warmup period in tap mode or with modifier-key bindings
- Requirements and limitations:
- Requires a Claude.ai account — does not work with API key, Bedrock, Vertex, or Foundry auth
- Microphone access required
- Does not work in remote/SSH sessions, WSL without WSLg, or devcontainers — the Claude Code process inside a container has no access to the host microphone; VS Code audio forwarding does not bridge this gap
- Supported on macOS, Linux, and Windows (WSL2 + WSLg)
- Audio processing: transcription runs on Anthropic servers; does not consume tokens or count toward usage limits
- Languages: 20+ supported (English, Spanish, French, Japanese, Chinese, and more)
- Source: Claude Code voice dictation docs