The ButterFilter, when you wish to use it for a 1-D signal (a numpy array with shape of (1, ), then you create it with f.e.: ButterFilter([1, None], sampling_freq, 1), with n_channels being the 1.
This now become a problem when you apply the filter on 1-D signals (such as are created after applying all of the Spatial Filtering). Since the 'zi' is defined with no.zeros(X, 1), which has a shape of f.e. (5, 1) and therefore scipy will itself spit out the "ValueError: object of too small depth for desired array".
One possible solution could be to np.squeeze it in line 92 of the filters.py in directory signal_processing. But perhaps there's something more elegant. Do you wish me to ask pull requests when I encounter things?
Dear Alex and Nikolai,
Something I came across when using your toolbox:
The ButterFilter, when you wish to use it for a 1-D signal (a numpy array with shape of (1, ), then you create it with f.e.:
ButterFilter([1, None], sampling_freq, 1), with n_channels being the 1.This now become a problem when you apply the filter on 1-D signals (such as are created after applying all of the Spatial Filtering). Since the 'zi' is defined with no.zeros(X, 1), which has a shape of f.e. (5, 1) and therefore scipy will itself spit out the "ValueError: object of too small depth for desired array".
One possible solution could be to np.squeeze it in line 92 of the filters.py in directory signal_processing. But perhaps there's something more elegant. Do you wish me to ask pull requests when I encounter things?