ValueError: all the input arrays must have same number of dimensions about acousticeventdetection HOT 3 CLOSED

I'm actually not sure what might cause this error. It seems like python_speech_features can't extract the spectrogram of certain files. Maybe there was a problem with opening the files and the 'sig' variable is empty. You could try to 'print sig.shape' before it's passed to 'getSpecFromSignal' to see if it's empty or None. If thats the case, we have to dig a bit deeper and figure out why psf can't open the files.

from acousticeventdetection.

Thanks for replying！
I've tried to print sig.shape, it turns out to be (176400, 2). I think it should be (176400,) after reading the code of "python_speech_features\sigproc.py", it seems that all the signal process functions assume the signal is one dimension, for example, there are two statements in sigproc.py:
slen = len(sig)
zeros = numpy.zeros((padlen - slen,)
padsignal = numpy.concatenate((sig,zeros))
the input of numpy.concatenate function must have the same dimension.
Thanks!

from acousticeventdetection.

So it's probably a stereo recording and you need to condense it to mono. This might be possible with PSF - if not, you can simply check if there is a second dimension and just remove it.

from acousticeventdetection.