Open-source speech recognition model for local transcription and subtitles
Overview
- Purpose
- Open-source speech recognition model for local transcription and subtitles
- Category
- Audio
- Developer
- OpenAI
- Official site
- https://github.com/openai/whisper (opens in a new tab)
Main features
- Recognizes and transcribes speech in many languages
- Detects the spoken language
- Translates speech into English
- Command-line output as text, SRT, VTT and other formats
Best for
Suited to developers comfortable with the command line or building speech recognition into their own workflows. For transcription with a graphical interface, consider Buzz, which uses local models.
Things to know
- Installation requires Python and FFmpeg, plus downloading model files
- Model size affects speed and computing requirements
- Recognition quality depends on recording quality, language and model choice
FAQ
- Is Whisper a web tool?
- The entry here is OpenAI's open-source project, usually installed and run on a computer or server.
- Can it generate subtitles?
- Yes. The official command line outputs SRT, VTT and other formats.