Live Whisper transcription, and Voice Assistant!

[Nikorasu]

This is my "real-time" implementation of Whisper, which sentence-by-sentence transcribes microphone audio to the terminal.
https://github.com/Nikorasu/LiveWhisper
Uses the sounddevice library, because my system has conflicts with with PyAudio..

I've also included my Voice Assistant project, which uses this Whisper implementation as a replacement for the standard speech-recognition library, (that I can't use cause it relies on PyAudio).
For now, the Assistant's abilities are simple: It can retrieve weather or Wikipedia searches, date and time info, and jokes..
New: It can now control any media-player on the system, long as you have echo-canceling: play, pause, resume, stop, previous, next, volume up, volume down, and what's playing, all should work, including variations on ways to say those.
Thanks to Whisper, it works really well! And I should be able to add more features as I figure them out.

If anyone has any suggestions to improve how I'm doing things, I'd love to hear it!
For example, I couldn't figure out how to send numpy audio data to directly to Whisper. But I think that may require altering Whisper somehow..

Anyway, hopefully this helps other people who have the same audio-library issues as I did! ;)

[JBX2060]

Wow, just what I was looking for!!

It'll be great if you add ChatGPT's API responses next!

[UmangRajpara13]

TalkGPT

[tumor6]

Epic work. !

[palladium123]

Love this! Speech Recognition didn't work for me as intended, but this did. With some modifications it works really well for live dictation, hardware specs notwithstanding...

[baibhavKumar1]

How to use this?