Thank you for your replies! They really helped me getting started. One of the things that author does is define a specific shape of the filter (I think to make a smoother transition?) with the following Matlab code:
nyquist = EEG.srate/2;
lower_filter_bound = 4; % Hz
upper_filter_bound = 10; % Hz
transition_width = 0.2;
filter_order = round(3*(EEG.srate/lower_filter_bound));
% create the filter shape
ffrequencies = [ 0 (1-transition_width)*lower_filter_bound lower_filter_bound upper_filter_bound (1+transition_width)*upper_filter_bound nyquist ]/nyquist;
idealresponse = [ 0 0 1 1 0 0 ];
filterweights = firls(filter_order,ffrequencies,idealresponse);
So ffrequencies defines the frequencies of interest, normalized to the nyquist frequency. And then the shape of the filter is defined by idealresponse. Is there a way to make a smoother transition (assuming this is the reason) like this in Julia?
Also the filter order here comes out to something like 192, this doesn’t work with Butterworth(192). What exactly is this value? And is it the same as the filter order for butterworth and the “order- n FIR filter” (sorry if this latter part may be beyond the scope of this discussion)