Gemini CLI voice input
Gemini CLI has a built-in voice mode, but it's experimental and off by default. Turn it on in settings and you get /voice and push-to-talk on the Space key. By default your recordings go to Google Cloud, and there's a setting to use local Whisper. This page is based on Gemini CLI's configuration docs. We didn't run it ourselves, and experimental settings can change, so check the docs if something doesn't match.
Turn it on
Add this to your Gemini CLI settings.json, in the user or project scope, then start a new session:
{
"experimental": {
"voiceMode": true,
"voice": {
"activationMode": "push-to-talk",
"backend": "whisper",
"whisperModel": "ggml-large-v3-turbo-q8_0.bin"
}
}
}
Only voiceMode is required. Leave out the voice block and you get the defaults: push-to-talk and the gemini-live backend.
The settings
| Setting | Default | Options |
|---|---|---|
| experimental.voiceMode | false | true, false |
| experimental.voice.activationMode | push-to-talk | push-to-talk, toggle |
| experimental.voice.backend | gemini-live | gemini-live, whisper |
| experimental.voice.whisperModel | ggml-base.en.bin | tiny.en, base.en, large-v3-turbo q5_0, q8_0 |
| experimental.voice.stopGracePeriodMs | 4000 | Milliseconds to wait for the final text |
From docs/reference/configuration.md in github.com/google-gemini/gemini-cli, checked 6 November 2026.
Things to watch for
- Language. The default Whisper model, and tiny.en, are English only. For other languages, use a large-v3-turbo model.
- Privacy. If you're dictating prompts about private code, choose the backend deliberately, since the default sends audio to Google Cloud.
- Permissions. On a Mac, your terminal app needs microphone access. Check System Settings → Privacy & Security → Microphone if nothing is recorded.
- Bugs. It's marked experimental. Gemini CLI's issue tracker has had voice bugs, such as a gemini-live connection closing with code 1000. Look there first if it fails.
Without the built-in mode
Voice in Gemini CLI only works inside Gemini CLI. If you move between it, Claude Code, Codex and an editor, a system-wide tool saves you from setting up each one. FnFlow is free and uses whisper.cpp with large-v3-turbo q8_0 on your Mac's GPU. Hold fn or right ⌘ in the terminal, talk and release. See dictation in Terminal, iTerm2 and Warp and how to speak a good prompt.
Questions
How do I enable voice in Gemini CLI?
Set experimental.voiceMode to true in your settings, then start a new session. That enables the /voice and /voice model commands.
Is Gemini CLI voice local?
Not by default. The default backend is gemini-live, and the docs say recordings are sent to Google Cloud for transcription. Set experimental.voice.backend to whisper for local transcription.
Which Whisper model should I pick?
The default is ggml-base.en.bin, and the options are tiny.en, base.en, large-v3-turbo q5_0 and large-v3-turbo q8_0. In our own whisper.cpp test on an M4 Pro, q8_0 matched the full model on 31 of 32 dictations, and q5_0 changed words in 11. Base struggled in earlier runs. If you have the disk space, use q8_0.
Do I still need the community extension?
Its author wrote that push-to-talk would need to be built into Gemini CLI itself. The built-in setting now offers it, so try that first.
Local Wispr Flow alternativeFnFlow vs VoiceInkVoice input for CursorHow 7 Mac dictation apps get you to your first dictationWhisper.cpp on Apple Silicon: which large-v3-turbo quantization to shipHow I built a local dictation appVoice dictation in Claude Code and the terminalWhy holding fn doesn't work for dictation on Mac (and how to fix it)How to run Whisper locally on an Apple Silicon MacSpokenly alternative for developersFnFlow vs EnviousWisprHow to dictate a good prompt for AI coding toolsVoice dictation in VS Code on a MacTeaching Whisper your jargon: initial prompt and replacementsWhy Whisper types "Thank you" on silence, and how to filter itPrivate dictation on a Mac at work: what actually leaves your machineVoice input for Codex CLI and the ChatGPT appVoice coding on a Mac: what actually worksWriting git commit messages by voiceDictating code review comments on GitHubDictation in IntelliJ IDEA, PyCharm and WebStorm on a MacDictation in Zed on a MacDictation in Terminal, iTerm2 and WarpWhich Whisper model to use on an Apple Silicon MacPush-to-talk dictation on a Mac: hold, toggle or streamSuperwhisper alternative for developersClaude Code voice mode: how to enable /voice and fix itWhisper apps: five ways to run Whisper locallyTalon voice coding vs a dictation appPush to talk on a MacDoes Claude have a voice mode?VoiceInk vs Wispr FlowVoiceInk vs SuperwhisperParakeet vs Whisper on a MacApple SpeechAnalyzer vs WhisperWhisper.cpp vs faster-whisper on a MacMlx-whisper vs whisper.cppGitHub Copilot voice inputVoice input in Google AntigravityVibe coding with voice input