ufal/whisper_streaming
A real-time streaming implementation of the Whisper speech recognition model for continuous speech-to-text transcription with 3.3 second latency.
Not currently ranked — collecting fresh signals.
star history
This project extends OpenAI’s Whisper model to support real-time streaming transcription and translation. It implements a local agreement policy with self-adaptive latency to enable continuous inference on unsegmented long-form audio. The system was designed for live transcription services and demonstrated at a multilingual conference.
Frequently asked
- What is ufal/whisper_streaming?
- A real-time streaming implementation of the Whisper speech recognition model for continuous speech-to-text transcription with 3.3 second latency.
- Is whisper_streaming open source?
- Yes — ufal/whisper_streaming is open source, released under the MIT license.
- What language is whisper_streaming written in?
- ufal/whisper_streaming is primarily written in Python.
- How popular is whisper_streaming?
- ufal/whisper_streaming has 3.6k stars on GitHub.
- Where can I find whisper_streaming?
- ufal/whisper_streaming is on GitHub at https://github.com/ufal/whisper_streaming.