Run time gradually becomes slower if a large number of augmented spectrgrams are being generated using a for loop #32
Description
When more than 1000 audio files are being fed to this tensorflow SpecAugment code using a python loop, the execution times gradually become very slow. Is there any way to solve it?