Calculate Time-varying PSD

Hi, I want to get the time-varying PSD.
If I do average=None for psd_welch, then the output would be (n_channels, n_freqs, n_segments), which is a time-varying PSD.
However, I want to make sure that, will the functions like mne.time_frequency.tfr_morlet,mne.time_frequency.tfr_multitaper or mne.time_frequency.tfr_stockwell give me the same result?

They won’t give the same result, but they should all give similar results whose differences are explained by the differences in underlying computations. IIRC the tfr_* functions will return objects with the same times as the original input, and this will be longer than the Welch output unless you use n_overlap=n_window - 1 (i.e., a 1-sample sliding window).

1 Like