What is the Fastest Whisper configuration

[dmarcus-wire]

As a data scientist, I want to test the fastest parameters of different Whisper model sizes (i.e. tiny, small, base, medium, large, turbo) overall. How fast can we get it to transcribe against input-samples.