CPU Port of OpenAI's Whisper Speech to Text (github.com)

Wow this is great!

I was planning to use the Python version and it’s really exciting to see a cleanly implementation in C++.

Great to see a C/C++ CPU-based implementation that runs on a variety of platforms. Most other ASRs seem to be either abandonware, a whole bunch of Python that only runs on PC, and/or solutions that send all your audio to someone's server.

I may be (probably am) being stupid, but do any of the benchmarks they publish actually compare it with the original PyTorch version?