Whisper (open source)

Whisper is OpenAI's open-source speech recognition model, published as the openai-whisper Python package. It runs on your own machine and transcribes audio files in many languages.

Website Docs Changelog Compare with others

Whisper (open source) at a glance

From official pages, checked Oct 4, 2026. Tap the arrow for each source.

Official SDKs: Python
Official package is openai-whisper on PyPI, requires Python 3.8 or later.
Open source: Yes, MIT License
The README states the code and model weights are released under the MIT License.

What you can build with Whisper (open source)

Whisper (open source)RecipeConfig

Transcribe audio on your own machine with open-source Whisper

Install openai-whisper and ffmpeg, then transcribe a local audio file with a short Python script. No API key, no upload.

Voice

Whisper (open source) at a glance

What you can build with Whisper (open source)

Transcribe audio on your own machine with open-source Whisper

Update timeline