Automatic_Speech_Recognition icon indicating copy to clipboard operation
Automatic_Speech_Recognition copied to clipboard

Multi-threading preprocess

Open RoyJames opened this issue 6 years ago • 0 comments

The preprocess (as well as converter) basically iterates and extracts features for each file independently. This can be easily multi-threaded with python on a multi-core machine or bigger clusters and get linear speedup. This can be beneficial especially when you want to experiment with preprocess hyperparameters. I can possibly do that (not sure when) if nobody has such plan yet.

RoyJames avatar May 17 '18 22:05 RoyJames