HumlaGlossary
What is Whisper (speech recognition)?
Whisper is OpenAI's open-source speech-recognition model, widely used to transcribe audio and runnable entirely on your own hardware.
- Free, no account
- Open source, MIT
- Apple Silicon, macOS 13 or later
In depth
What it means
Because the model weights are open, Whisper can run locally — including on Apple Silicon GPUs — instead of calling a paid cloud API.
Humla bundles a local Whisper engine (a quantised large-v3-turbo model) and a Norwegian-tuned NB Whisper model, so on-device transcription is free after a one-time download.
Related
Where this comes up
Keep your meetings on your Mac.
Free and open source. Apple Silicon, macOS 13 or later.