Whisper
Open-source speech recognition model from OpenAI
Whisper is a general-purpose speech recognition model from OpenAI that transcribes audio, translates speech and identifies languages. Its repository describes it as trained on diverse audio with large-scale weak supervision, and both code and model weights are released under the MIT License. Six model sizes are offered, from tiny at 39 million parameters to large at 1,550 million, plus a turbo model that is an optimized version of large-v3. Smaller models have English-only variants. It runs locally through a Python package and needs ffmpeg installed, with required GPU memory ranging from about 1 GB to about 10 GB depending on the model.
- Filed under
- Audio & Music
- Pricing
- Free
- Platforms
- desktop · api
- Last updated
- November 10, 2026
What Whisper does
- Multilingual transcription. Recognizes speech in many languages; accuracy varies widely by language.
- Speech translation. Translates speech to English using multilingual models.
- Language identification. Detects the spoken language of audio.
- Six model sizes. Tiny to large plus turbo, trading speed against accuracy.
- English-only variants. The tiny, base, small and medium sizes have .en versions.
Who it’s for
- Developers running transcription on their own hardware
- Researchers needing an open speech model
- Projects that need MIT-licensed speech recognition
Whisper pricing
Free open-source (MIT License); you supply your own compute
As listed on November 9, 2026. Check github.com for current prices.
Things to know
- Turbo model is not trained for translation tasks
- Large model needs about 10 GB of VRAM
- Requires Python, ffmpeg and command-line or code use; no GUI in the repository
Quick answers
- What is Whisper?
- Whisper is a general-purpose speech recognition model from OpenAI that transcribes audio, translates speech and identifies languages.
- Is Whisper free?
- Yes. Whisper is free to use. (As listed on November 9, 2026.)
- Who makes Whisper?
- Whisper is made by OpenAI.
- Where can you use Whisper?
- Whisper is available on desktop (apps, editor plugins or browser extensions) and an API for developers.
- What are some alternatives to Whisper?
- Other tools filed in the same section include Bark, Happy Scribe and Krisp.
Tags
#speech-recognition#transcription#translation#open-source#mit-license
See also
Cross-referenced entriesBark
Open-source transformer text-to-audio model by Suno
text-to-audioopen-sourcespeech
Happy Scribe
AI transcription, subtitles and meeting notes
transcriptionsubtitlesnotetaker
Krisp
AI noise cancellation, meeting notes and accent conversion
noise-cancellationmeeting-notesaccent-conversion
AudioCraft
Meta's open-source library for generative audio research
musicgenaudiogenencodec