Spectrograms

Spectrograms display the signal’s frequency–time decomposition.

Select the signal from the top-left drop-down (only signals with sampling rates of 32 Hz or more are listed here). Then press Spectrogram to generate a view, where the x-axis is each epoch across the recording, the y-axis is frequency (Hz) and the z-axis (color heatmap) is spectral power.

You can right-click on the plot to either copy the image to the clip-board, or save it as a file.

Spectrogram parameters

Frequency ranges are customizable: here we increase the maximum frequency to 80 Hz (and repress Spectrogram) to generate this image:

Sometimes the dynamic range of the z-axis is restricted, due to outliers (e.g. very high amplitude signal artifact at the end of the recording). In this case, it can help to winsorize the range of z-values plotted (i.e. clipping to the n-th percentile). In this example we winsorize the same spectrogram as at the top of this page, with a value of 20% (0.2):

Masked/gapped recordings

If the EDF contains masked epochs, the plot should show that as follows, e.g. here with epochs restricted to N2 only:

Hjorth plots

A generic alternative to a spectrogram is a Hjorth plot; here, the Y-axis shows magnitude (first Hjorth parameter), and the top/bottom colors indicate the second/third Hjorth statistics of mobility and complexity respectively:

As a simple representation of signal amplitude, these plots can be useful to show structure/artifact for signals where (unlike the EEG) a typical spectrogram may be less useful.