I have troubles testing Automatic Speech Recognition models on the Hub.
For example, if I visit openai/whisper-medium · Hugging Face with my browser and I click on the Realtime Speech Recognition button and grant audio input permission in the browser, I can’t get past model loading:
This happens both with Firefox 108 and Edge 108 on Linux and Windows. What could the problem be here? Is the ASR widget working for you? Recording from browser + computing (so, with no real time recognition) works fine instead. Thanks!