This python script implements stretching or shrinking audio data without changing the pitch. Given an input wav file, it stretches(or shrinks) it and saves to a new output file.
The audio is first converted to frequency domain resulting in a short-term fourier transform (STFT). The STFT is then transformed according to The Phase Vocoder algorithm, and ultitmately converted back to time domain. The stretch is controlled by the stretch factor, the new duration being old duration times the stretch factor.
The file can also be imported as a library to call the stretch function. The functionality is inside stretch() function. I wrote a loop-less version of this function as stretch_wo_loop() in order boost speed, but it just led to less a readable function without much change is speed.
The command line usage for running the script is:
usage: speed up or speed down the audio without changing the pitch
[-h] -i IN_FILE [-o OUT_FILE] -f FACTOR [-n NFFT]
optional arguments:
-h, --help show this help message and exit
-i IN_FILE, --in-file IN_FILE
path to the input wav file
-o OUT_FILE, --out-file OUT_FILE
path of the output file
-f FACTOR, --factor FACTOR
factor by which to shrink or dilate time. if FACTOR <
1.0, audio will be sped up, otherwise sped down
-n NFFT, --nfft NFFT num of FFT bins to use