Analysis updated 2026-05-18
Dictate messages, notes, or comments hands free into any Mac app without switching windows.
Speak prompts into AI chat tools or search boxes instead of typing.
Clean up voice transcripts automatically using your own OpenAI or Anthropic compatible API key.
| bytonylee/mac-whisper | oomol-lab/closeup | arinltte/luce | |
|---|---|---|---|
| Stars | 30 | 30 | 31 |
| Language | Swift | Swift | Swift |
| Setup difficulty | easy | easy | easy |
| Complexity | 2/5 | 2/5 | 1/5 |
| Audience | general | general | general |
Figures from each repo's GitHub metadata at analysis time.
A downloadable DMG installer is provided, building from source needs macOS 26+ and Xcode command line tools.
Mac Whisper is a free menu bar app for Mac computers that lets you dictate text into any app on your screen using push to talk. You hold down the Fn key, speak, and let go, and the words you said are pasted directly into whatever text field is currently active, whether that is a chat app, a notes app, a search box, or anywhere else your cursor happens to be. While you are speaking, a small floating panel shows the live transcript so you can see what is being recognized. The app can also send the transcript through an AI cleanup step before pasting, which fixes recognition mistakes and awkward phrasing. This step is optional, and if the AI request fails for any reason, the app simply pastes the raw, unedited transcript instead so dictation never gets stuck. Under the hood, the app watches for the Fn or Globe key using low level keyboard input monitoring, records audio with Apple's built-in Speech framework, and pastes the result by simulating a copy and paste action, restoring whatever was on the clipboard beforehand. It prefers the Mac's built-in microphone during dictation so that Bluetooth headphones are not forced into a lower quality call mode. Speech recognition supports English, Korean, Chinese, and Japanese. To use the optional AI cleanup, you supply your own API key for an OpenAI-compatible or Anthropic-compatible service through an environment variable, so the key is never stored in the app's saved settings. The app also explains its permission needs plainly: microphone and speech access to transcribe what you say, input monitoring only to detect the Fn key press, and accessibility access only to paste text into the app you are using. It does not write your transcripts to any diagnostic log. You can download a ready-made installer from the project's releases page, or build the app yourself from source using the included build script, which requires macOS 26 or later and Xcode's command line tools. The project is released under the MIT license.
A free Mac menu bar app that lets you dictate text into any app by holding the Fn key, with optional AI cleanup of the transcript.
Mainly Swift. The stack also includes Swift, macOS Speech framework, HID.
MIT license: free to use, modify, and distribute, including for commercial purposes, as long as the copyright notice is kept.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly general.
This repo across BitVibe Labs
Verify against the repo before relying on details.