No API key
Nothing to request, nothing to bill, nothing to paste into a settings pane. Install it and talk.
The voice input terminals never had
Hold a key, say it, and it's typed.
Hold the right Option key, talk, let go. The text lands wherever your cursor already is — a shell, an editor, a browser field. Recognition happens on your Mac, so there's no key to paste and nothing to upload.
Download for MacmacOS v1.7.5 · DMG · free, always- No API key
- Runs on-device
- Free, always
- Any text field
Terminals can listen too.
ready
Last login: Fri Oct 24 22:51:08 on ttys001
chris@MacBook-Pro voice-doggo % claude
›_ Claude Code is ready. Describe what you want to build.
›
Fix this page for me, then run the tests.
On it. I'll run the tests afterwards and tell you how they went.
Try the right ⌥ — same key, every app
four things, that's all
No setup, no second-guessing.
Just the thing you wanted to say.
Your voice stays on the Mac
The model is bundled and runs locally. No audio is uploaded, which is also why it keeps working with the Wi-Fi off.
Free, and not the trial kind
No subscription, no minutes, no word count. Free now, and free the next time you open it.
Any text field counts
Terminal, editor, browser, desktop agent, an SSH session three hops away. Wherever the cursor is, that's where the words go.
how it compares
Why this one?
Plenty of AI tools ship voice input now. The catch is where it lives: inside that product's own chat box, usually with the audio going to a server first. Voice Doggo sits a layer lower — it types into whatever window has focus, and the audio never leaves the machine.
See the full comparison
| Voice Doggo | Voice built into an app | Cloud speech API | |
|---|---|---|---|
| Where it types | Any focused window, terminals included | That app's own input box | Wherever you wire it up |
| Getting started | Install, grant two permissions | Open the app | Request a key, write the integration |
| Where the audio goes | Stays on your Mac | Usually to a server | To a server, by design |
| With the network off | Still works | Depends on the product | Doesn't work |
| Cost | Free | Depends on the plan | Metered or subscription |
Built-in voice features change from release to release. This compares the shape of each approach, not any one product on any one day — check the vendor's own docs before you rely on a row.
macOS 13+ · Apple Silicon (M-series) only
Type less. Make more.
faq
Before you download,
the things people ask.
Do I need an API key?
No. There is no key to request, no account to create and nothing to paste into a settings pane. Install it, grant Accessibility and Input Monitoring, and start talking.
Is it actually free?
Yes. No subscription, no per-minute billing, no word quota, no paid tier holding the good part back.
Which apps does it work in?
Any macOS app with a text field. It delivers the transcript as synthetic keyboard input to the focused window, so Terminal, iTerm2, Cursor, Claude Code, VS Code, browsers, docs and chat apps all work the same way — including shells running over SSH or inside tmux.
Does my audio leave the Mac?
No. Recognition runs on-device with a bundled offline model (FunASR SenseVoice). Nothing is uploaded to a speech API, and dictation keeps working with the network off.
Which Macs are supported?
Apple Silicon (M-series) Macs running macOS 13 or later. Intel Macs are not supported: recognition runs locally, and on Intel it is too slow to use.
Does it understand English?
Yes. The on-device model detects the spoken language itself and handles English, Chinese, Cantonese, Japanese and Korean. What it cannot do is learn custom vocabulary — the model has no hotword support, so camelCase identifiers and abbreviations are where you should expect to fix things by hand. Describing intent in plain sentences works better than spelling out symbols.
I hold the key in my terminal and nothing happens.
Check Secure Keyboard Entry first. With it on, macOS stops delivering key events to third-party event taps, so the app never sees the key you are holding and fails silently. Terminal.app has the toggle in its Terminal menu; iTerm2 keeps it under Preferences → General.
further reading
On talking to a terminal
-
Voice Input in the Mac Terminal (Not the
sayCommand) Why built-in Dictation falls flat in terminal emulators, and the approaches that actually land text. - Mac Dictation Not Working in Terminal The causes, ordered by likelihood, starting with Secure Keyboard Entry. Each one with a way to confirm it.
- Offline voice input for Claude Code, including over SSH What changes when the recognition runs on your machine instead of streaming to a server.