Open gabrilator opened 1 year ago
Hey!
So I tried loading the model in Banana dev, and the performance is very slow still. It takes around 10 minutes to process a 30 second audio.
I wonder if there are any key factors to consider in terms of performance. I see that the huggingface app has great performance!
Cheers, gabrilator
Hey @gabrilator - it runs quite quickly for me on a Mac M1 on CPU. Could you share your code for benchmarking? Happy to take a look into why it's so slow!
Hey y'all!
Been doing some tests of the diarization pipeline, and unfortunately I'm getting very slow processing times.
Granted, I'm testing it from my Mac M1, but when I transcribe stuff with the "barebones" Whisper I do get faster processing times.
As of right now, I am using an API KEY generated from a free Huggingface account.
Is this slow performance expected? Is there a way to check any potential performance bottlenecks?
Thank you!