Hi, I am using the wav2world function for the sp so that I can further calculate mcep with SPTK.
I have a similar problem as #84.
I am using fs=16000, frame_period=16.0, fft_size=1024.
I suddenly found that if I apply the function to an audio with exactly 28160 data points, i.e., (16000*0.016)*110,
the output size of sp is (111, 513) while I expect 110.
Does PyWorld start the first frame at the first data point or does it pad something before or after?
May I also ask what is you suggestion if I really want to have 110 output frames?