A newbie's question on Fourier using freqanalysis_mtmfft

[Nathan Weisz]

> Lastly, an even more general question: What do you think would be handier: (1) Compute the frequency spectrum over 60 seconds, to get a high frequency resolution or (2) to cut the signal into e.g. 4 second pieces and to average all spectra to increase the signal to noise ratio?

as suggested before do not average the spectra, but the time series. after that you do your fourier stuff (with cfg.taper='hanning' as ingrid suggested).

beware to move the window in steps which are exact multiples of your steady-state frequency cycle (e.g. 100 ms for 10 Hz). when you calculate the amount of sampling points you need to move your window, then first "think" in latencies (for 10 Hz e.g: 0, 100, 200 etc ms, if you move 1 cycle), after that convert to sampling points (e.g. using nearest.m). this is very important, because when you first "think" in sampling points then you might be missing your desired time point each time just a little bit, however the inaccuracy accumulates with each moving of your window. in the end you see no steady-state at all. this mistake did cost me and a phd student a couple of days to figure out while doing auditory steady-state ...

good luck,
nathan





----------------------------------
The aim of this list is to facilitate the discussion between users of the FieldTrip  toolbox, to share experiences and to discuss new ideas for MEG and EEG analysis. See also http://listserv.surfnet.nl/archives/fieldtrip.html and http://www.ru.nl/neuroimaging/fieldtrip.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20091222/16705aa6/attachment.html>