https://github.com/antirez/voxtral.c/blob/main/python_simple...

Thanks a lot! Let me see if I can get it to work with pure numpy while keeping the input audio part streaming.