Hi all, I am trying to write a praat script to do preprocessing on hundreds of speach samples. I need to extract speech features from each sample and feed these as imputs into a feed-forward neural network. I have already constructed the network using math-lab. However, learing to script in praat is proving to be quite a challenge given my time constraints.

Some of my samples are 0.01 to 0.03 seconds in length, I was looking at standardising the duration for all samples using Pitch Synchronous OverLap-Add(PSOLA). However this will be very tedious if I were to do this for every sample.

Is there any script that can read in all of my files and perform the operations in a batch mode? Any guidance will be surelly appreaciated.

Regards.