GitHub radar
Audio8 ASR Infinite: an infinite stream of speech recognition models
Edge0 has published a long-term stable streaming ASR model which prevents the long audio from being messy at the end of the recording.
Edge0 has published a long-term stable streaming ASR model which prevents the long audio from being messy at the end of the recording. Audio8 ASR Infinite is an online speech recognition model developed by Edge0 which converts the audio signal into corresponding texts in real time. With the help of the streaming model, it can process data with a high decoding speed (12.5Hz). At the same time, it also supports the adjustment of delay time (240ms–560ms) and can recognize both Chinese and English languages. This model is relatively small, with only about 4 billion parameters. However, since it is not a general application, we can only deploy it on our own server via Docker to run it as a lightweight streaming service. We also provide a web demonstration.
Why a vibe-coder should care
This model is designed for long audio transcription such as interview, live streaming, meeting, etc., so that the long audio won’t cause trouble when transcribing. With low latency (less than 1s), you can see the subtitles nearly real time instead of missing out the rest of your speech. Unfortunately, it requires you to build a Docker environment rather than simply install an application.
How to install
Copy this and send it to your agent — Claude Code, Codex, any of them:
Set up the Audio8 ASR Infinite model from Edge0 for me: https://huggingface.co/Edge0/Audio8-ASR-Infinite — follow their Docker setup, launch the web demo, and show me how it transcribes my speech in real time.
A regular laptop with 8 GB of RAM or more can hold the model — the weights take about 8 GB. For real-time transcription a GPU helps a lot: the developer's own examples run it on one, and on a CPU alone the response lags noticeably.
Open on Hugging Face▌ More finds