Best way to implement streaming application?

[9throok]

Hi @sanchit-gandhi, I was working on and around whisper and its applications for quite some time. Although its been a while since I last touched this topic, there have been so much development after distil-whisper and V3 were announced.

I come here after visiting resources, forums, discussions and threads like these:

• Compatibility with CTranslate2 / faster-whisper #3
• example using microphone #4
• small uses more memory (but is faster) than medium (ONNX quantized) #53
• https://github.com/SYSTRAN/faster-whisper#usage

If, lets say I were to build some thing like this, what will be the best way to do it, considering only short audio chunks?

It would be great help if you could share your insights on it.