Frequency analysis

✉️ Mailing List Archive (read-only)
1

External Email - Use Caution

Dear MNE team,

First of all, thank you for your hard work.

I am comparing different methods for EEG frequency analysis, and I have the following question/suggestions:

  * in "psd_array_multitaper" the user can decide to normalize the PSD by N or N*Fs, however in "psd_array_welch" such choice is not available. Is this the wanted behaviour ? "psd_array_welch" is normalize with N*Fs (default of scipy spectrogram I believe)
  * In "psd_array_welch", the type of windowing is restrained to 'hamming' windows. Would it be possible to add an argument to choose it ?
  * I am not very familiar with stockwell transform, but is it possible to get a spectrogram out of it ? I wanted to compare it with the other methods.

Cheers,

Bastien

2

External Email - Use Caution

   - in "psd_array_multitaper" the user can decide to normalize the PSD
   by N or N*Fs, however in "psd_array_welch" such choice is not available. Is
   this the wanted behaviour ? "psd_array_welch" is normalize with N*Fs
   (default of scipy spectrogram I believe)

We should probably just use the Welch-style normalization everywhere. It

should at least be the default. Feel free to open an MNE-Python GitHub
issue and we can decide how to unify these.

   - In "psd_array_welch", the type of windowing is restrained to
   'hamming' windows. Would it be possible to add an argument to choose it ?

Sure, feel free to make a PR to add this.

   - I am not very familiar with stockwell transform, but is it possible
   to get a spectrogram out of it ? I wanted to compare it with the other
   methods.

You can get a time-frequency representation of the data from the Stockwell

transform. I am not too familiar with it either, but Its Wikipedia page
<https://en.wikipedia.org/wiki/S_transform&gt; says it's a generalization of
the STFT.

Eric
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20190830/34cb2f10/attachment.html