Open
Description
This might be a bit tricky, but for lots of users in vivo, it seems that noise levels can vary quite drastically over time, depending on the state of the brain (awake, anaesthetized, sleep, ...). The problem with such states is that if we are doing a single estimation of the noise levels, it might drastically affect the detection thresholds during the recording. The idea would thus be to allow a running estimate of the noise levels, and offer a whitening or a z_score normalization that will depend only on the recent activity.