speechbrain
Real-Time ASR with Device Capture using torchaudio StreamReader
Describe the bug
I made a demo of using StreamingASR with live device input using torchaudio StreamReader. The setup is a little tedious, and it would be nice to integrate this as an official feature to abstract threading and buffering away from the user.
Of course this can be applied to any model or pipeline, provided there are streaming capabilities.
Expected behaviour
Streaming ASR from device
To Reproduce
No response
Environment Details
No response
Relevant Log Output
Additional Context
No response
@pplantinga @Adel-Moumen, this could be our first pipeline example, what do you think? I'm wondering if we could put some sore of voice activity detection in top of this to make it a full proper pipeline.
I'd be happy to tackle the issue if no one's claimed it yet
I would be happy having you on board. If you are willing to push this feature, and make it speechbrain-like, we can team up on that. What do you think? :)
