Dictate prompts with voice in Grok Build

Grok Build can dictate into the prompt with Ctrl+Space / F8, or with /voice. The three [ui] keys below tune keybinds, hold vs toggle, and speech-to-text language. Official: Settings reference.

Keys

Setting	Default	Role
voice_keybind_enabled	true	Enable Ctrl+Space / F8 (/voice still works when this is false)
voice_capture_mode	hold	hold or toggle
voice_stt_language	en (or [voice].language)	Language code or auto

User config example

[ui]
voice_keybind_enabled = true
voice_capture_mode = "toggle"
voice_stt_language = "en"

Path: ~/.grok/config.toml or $GROK_HOME/config.toml. Open /settings to change UI options without hand-editing.

Capture modes

• hold — press and hold Ctrl+Space or F8 while you speak; release to stop.
• toggle — press once to start dictation, press again to stop.

Use toggle when you need both hands free for notes. Use hold when you want dictation only while the key is down.

Language

Set voice_stt_language to a language code (en, es, …) or auto. When unset, docs fall back to en / [voice].language.

Disable keybinds only

[ui]
voice_keybind_enabled = false

/voice remains available after keybinds are off, so you can still dictate from the command palette path without Ctrl+Space / F8 colliding with the terminal.

Pitfalls

• Expecting /voice to disappear when voice_keybind_enabled = false — the slash command stays.
• Putting these keys in project .grok/config.toml — use user config for [ui].
• Mixing this with the Voice API / speech-to-speech product — those are separate how-tos under Voice.