Hi, what would be the easiest way to substitute current MFCC feature extraction method for a different one? Would changing the samples_to_mfccs function from util/feeding.py be enough? I’m still getting a bit lost when it comes to large projects. Thank you in advance
It should be enough.
Are you referring to the training process? I would like to change the MFCC extraction method only during decoding. Is there any way to do this? Please suggest.