Voice dictation for macOS · everything runs on your Mac
Speak. The text appears
right at your cursor.
Hold a key, speak, let go. Built for people in tech who mix English terms into Brazilian Portuguese, without sending their voice to the cloud.
Subi o deploy no Kubernetes e abri um PR no GitHub. Falta só o review do pipeline de CI/CD.
HOW IT WORKS
Hold, speak, let go.
No window to open, no button to click. Falae waits in the menu bar for your shortcut.
01Hold your shortcut
You pick the key on first launch. Hold it to talk, or tap once to start and again to stop.
02Speak normally
Portuguese with the English terms you already use at work.
03Let go
The text appears at the cursor, in any app: chat, email, terminal, code editor.
VOCABULARY
Your jargon, spelled your way.
Add a name or term and, if you like, how recognition tends to get it wrong. Falae fixes those words after recognition, and never touches an ordinary Portuguese word.
In Quality mode your terms are also given to the model as context, to help it get the jargon right.
| YOU SAY | → | FALAE WRITES |
|---|---|---|
| cube control | kubectlyour term | |
| Open AI | OpenAIcompact form | |
| CICD | CI/CDcompact form | |
| post gress | Postgresyour term | |
| rede | redeordinary word: left alone |
FEATURES
Made for dictating all day.
Fast or Quality
Fast answers instantly. Quality uses your terms as context and needs a Mac with 16 GB of memory or more.
Profiles
Keep the text as you said it, or let Apple Intelligence clean it up, rewrite it or translate it. One shortcut per profile.
In any app
Works wherever there is a cursor: chat, email, terminal, code editor, documents.
Model changes without gaps
While a new model downloads, dictation keeps using the current one.
Ready to dictate?
A check shows what is missing (microphone, Accessibility, model, shortcut) and fixes it in one click.
Seven languages
Interface in English, Portuguese, German, Spanish, French, Japanese and Chinese.
PRIVACY
Your voice never leaves your Mac.
No account, no cloud, no server listening.
Recognition on your Mac
Audio and text stay on your computer. Once a model is downloaded, dictation works offline.
Network only for the essentials
Each speech model downloads once, from Hugging Face, and the app checks for updates. Nothing else.
Rewriting stays local too
Profiles that clean up or translate use Apple Intelligence, on-device.
FAQ
Frequently asked questions
Which Macs are supported?
Apple Silicon Macs with macOS 26.2 or later. Quality mode needs 16 GB of memory or more; on smaller Macs, Falae uses a lighter model.
Do I need internet?
Only to download each speech model the first time and to check for updates. After that, dictation works without a connection.
Does it work in English and other languages?
Yes. Falae is built for Brazilian Portuguese with English terms, and you choose the spoken language in Settings. For other languages, there are multilingual models.
How accurate is it?
We are still measuring. Tests so far used sentences read by one speaker; spontaneous speech and other accents are not benchmarked yet. We would rather not promise numbers before testing with more voices.
Which permissions does it ask for?
Microphone, to hear you, and Accessibility, for the global shortcut and to type at the cursor. Screen Recording is optional: only profiles that use what is on screen as context need it.