spikeinterface
spikeinterface copied to clipboard
SpikeInterface
Remove artifact function with explicit timestamps
RemoveArtifactsRecording takes as input list_triggers, which are in frames. On the other hand, ms_before and ms_after are in milliseconds. This is converted to frames using the sampling rate later. As a result, one cannot make use of ms_before and ms_after if one uses explicit timestamps (perhaps because there are missing samples in the recording). @alejoe91 @samuelgarcia what do you think about changing this such that get_times is used to infer the conversion from ms to frames? This means the padding must be computed separately for each trigger, so the performance may take a hit. Alternatively, could make the user specify frames_before and frames_after, which are computed manually before calling remove artifact.
oops, sorry I thought this was the one about masking out single samples. I don't think there has been a PR to address this - I had opened it to ask you guys your thoughts. what do you think - specify frames before and after, or keep the ms before and after and just infer the number of frames it corresponds to with the explicit timestamps?
@khl02007 I would keep ms_before and ms_after and add optionally frames_before and frames_after. If the latter are givem, the ms_* params are ignored. @khl02007 does that sound ok?
By the way, happy to make these changes and submit a PR if you are OK with that
@alejoe91 Did this idea never pan out? Does RemoveArtifactsRecording now have that functionality?
