I wanted voice-to-text on macOS that didn't send my audio anywhere. Almost all local AI models have very noticeable differences to leading cloud models, which makes running them impractical if you're not willing to accept quality degradation. OpenAI's Whisper model seems to be the one exception to this, so I was surprised that there was no super easy to install way to get this going.
I made whisper-hotkey so that you can open your Terminal, paste in one command, and have perfect transcription running—and you never have to set up the keyboard shortcut or the ability for it to open on launch.
To install, open Terminal (search for "Terminal" in Spotlight) and run:
pip install whisper-hotkey && whisper-hotkey install