OBS Python Whisper4Streamers 1.1.1

This script is an alternate implementation of whisper online for OBS, providing most of the options available in CLI to a visual interface. It allows nearly real-time transcription of audio from any source available on OBS as long as it is monitoring audio (the source doesn't need to have an output).

For the script to run, you will need to install these Python modules (a mod of Whisper online is included with my script; no need to install it):

- librosa
- faster_whisper
- ctranslate2
- torch
- numpy

To use the GPU option, you will need to install
(It's not necessary to have an NVIDIA GPU to run the script, but if you do, it can be faster)

cuBLAS or cuDNN 9 for CUDA (depends on your NVIDIA card; for Radeon, only CPU is available so far)

https://developer.nvidia.com/cudnn-archive (win/mac)
https://developer.nvidia.com/cublas (Linux)

CUDA 12 is recommended but not needed. I could run on a previous version, but going too far back is probably not ideal.

PS: All codes are under MIT or BSD license and equivalent like librosa's ISC.