Dictate text into any Mac app, like notes, browsers, or messaging apps, by holding a hotkey.
Get speech turned into text in Russian, English, or about 100 other languages without an internet connection after setup.
Optionally clean up spoken text automatically to fix grammar and remove filler words.
Replace a paid cloud dictation subscription with a free, private, local alternative.
| neurosonya/my-whisper | 2arons/llm-cli | abe238/claude-video-plus | |
|---|---|---|---|
| Stars | 11 | 11 | 11 |
| Language | Python | Python | Python |
| Setup difficulty | moderate | easy | easy |
| Complexity | 2/5 | 2/5 | 3/5 |
| Audience | general | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires an Apple Silicon Mac (M1 or newer) and Python 3.10+, several macOS permissions must be granted manually.
My Whisper is a free tool for Mac that lets you dictate text into any app just by talking, similar to the paid service Wispr Flow, but everything runs locally on your own computer instead of in the cloud. You hold down the Option key while speaking in any app such as Telegram, a browser, notes, or email, let go, and within a couple of seconds the recognized text appears right where your cursor was. The key difference from cloud based alternatives is privacy and cost: your voice recording never leaves your Mac, there is no subscription to pay, and the internet is only needed once, to download the recognition models. Under the hood, it uses OpenAI's Whisper speech recognition model, run through a version optimized for Apple's own chips, to turn speech into punctuated text in Russian, English, and around 100 other languages. Optionally, it can also use a small local AI model called Gemma 3, run through a tool called Ollama, to clean up the raw transcription by fixing misheard words, correcting grammar, and removing filler sounds like um. This only works on Apple Silicon Macs (the M1 through M4 chip families), not on Intel Macs or Windows. You need Python 3.10 or newer, about 2 gigabytes of free disk space for the speech model, an extra 3 gigabytes if you want the optional text cleanup feature, and roughly 15 minutes to set everything up. Installation involves downloading the project files, double clicking an install script that sets up the environment and downloads the model, then double clicking a start script that adds a small icon to your Mac's menu bar. You then need to grant a few macOS permissions covering microphone access, the ability to insert text into other apps, and the ability to detect your held down hotkey. The README also covers optional automatic startup when your Mac boots, and includes a troubleshooting section for common permission related issues. The project is released under the MIT license, meaning anyone can use, modify, and share it freely. It was built by its creator with help from Claude Code, and would appeal to Mac users who want a private, free voice dictation tool without depending on cloud services or subscriptions.
A free, private voice dictation tool for Apple Silicon Macs that types out your speech into any app without sending audio to the cloud.
Mainly Python. The stack also includes Python, Whisper, mlx-whisper.
MIT license: use freely for any purpose, including commercial use, as long as you keep the copyright notice.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly general.
This repo across BitVibe Labs
Don't trust strangers blindly. Verify against the repo.