Why extract waveform feature with int64 in preprocess.py ?
mindmapper15 opened this issue · comments
If I correct, the maximum value of extracted waveform data can have is 65535(16bit MOL).
But the extracted waveform feature are converted and saved with np.int64 format, which leads to consuming more memory size and using more storage.
Are there any reason to do so?
For instance, higher bits(24~32bits) sampling supports?