Description
I have code that resamples many audio files using scipy.signal.resample. The code uses much more memory than I would expect to the point of freezing the OS. I've managed to simplify my code to a simple test case. This uses ~8Gb RAM after about 6 iterations but I would expect it should use much less memory than this.
I am using inputs of varying sizes, which seems to be significant. If I use an audio input of constant size there doesn't seem to be a problem.
from scipy.signal import resample
import numpy as np
def Test():
shps = range(11473744, 1000, -2)
for i, shp in enumerate(shps):
signal = np.random.normal(size=(shp, 1))
print (i, signal.shape)
targetSamples = signal.shape[0]*44100//48000
signal2 = resample(signal, targetSamples, axis=0)
print (signal2.shape)
del signal, signal2
if __name__=="__main__":
Test()
I might have missed something obvious, so this might not even be a bug!
- Scipy 1.4.1
- Numpy 1.18.4
- Python sys.version_info(major=3, minor=6, micro=9, releaselevel='final', serial=0)