Picture published by the creator on Hugging Face.
Transcribe speech live, continuously, with no length limit
Made by Edge0
View profile on Hugging Face (opens in a new tab)Audio8 ASR Infinite is a speech-to-text model built for real-time, unlimited-length transcription rather than processing a finished audio file after the fact.
Most streaming transcription models drift out of sync or need to be restarted for very long audio; this one uses a "rolling" memory technique that keeps both memory use and response delay constant even running 24 hours a day. It supports Chinese and English, lets you choose how often it emits new text (roughly 6 to 12.5 times per second) and how much delay to trade for accuracy, and includes logic to tell a thinking pause or a stutter apart from someone actually finishing what they were saying.
You could use it to…
- Transcribe speech live with no length limit, 24/7
- Build a captioning feature that never drifts out of sync
- Tell a thinking pause apart from someone finishing a sentence
Can I use this?
You'll need A GPU and the transformers library, or vLLM for 24/7 use · Setup needed
Worth knowing This is a preview release described by the creator as delivering only the transcription base, with a further "realtime semantic perception" capability still being built; accuracy figures in the model card are the creator's own reported error rates. Apache-2.0 licence. BuildTube has not run or verified this model.
Why it's here
Trending on Hugging Face
- 2,023 people have liked it on Hugging Face.
- It was downloaded 27,228 times in the last 30 days.
Numbers from the snapshot taken 1 October 2026; not refreshed since.
Behind it
See the code on Hugging FaceBuildTube has not run or verified this project. Everything above is written from what the creator published.
Made this?