reset to only audio processing

reset process from cpu to gpu

added missing dependency in requirements.txt and upgraded to whisper-v3

fixed processing through link or audio bytes

added possibility for video processing via link

initial commit