Hi, can do I a time-frequency decomposition on raw data?

I know I can do such tf decompostion on epoch data with tfr_morlet, can I do the same thing on raw data? After the decomposition, I would like to do epoching on the tf data which would end up with 4-D data instead of 3-D data during the normal epoching operation.

Hello @Long,

to my knowledge, this is currently not supported in MNE-Python.
I’m tagging @agramfort, who might be able to give you some more advice here.

Good luck,

it’s indeed not supported now. You can make a single long epochs to fake this at this point.

out of curiosity why do you need this? and why epochs are not ok for you?


Thanks for letting me know this. The reason I want to do tf decomposition on raw data is I have very short inter-trial interval. If I do tf decomposition on epoch data, there will be short artifact, or error at the beginning or end of the tf representation. So I think it would be better to do tf decomposition on the whole raw data, then I can cut the tf representation into epochs.

I would do the TF on extended epochs and crop the edges of TF objects


1 Like

Yes, that will do. Thanks for the advise.