For online experiments, I added causal filtering to RtEpochs (using lfilter in _process_raw_buffer), which circumvents the need for filtering on the segmented data (and needing longer epochs or risking edge artifacts).
Is there interest to include this in MNE? Then I would tidy up the code and create a PR.
Current limitation of my code is that filter delay is not compensated/events are not shifted, so users should use minimal-phase or IIR filters.
For online experiments, I added causal filtering to
RtEpochs(usinglfilterin_process_raw_buffer), which circumvents the need for filtering on the segmented data (and needing longer epochs or risking edge artifacts).Is there interest to include this in MNE? Then I would tidy up the code and create a PR.
Current limitation of my code is that filter delay is not compensated/events are not shifted, so users should use minimal-phase or IIR filters.