[Eeglablist] Problem with inverse transform

[Makoto Miyakoshi]

Hi Franzi,

By the way, I found that the Hann window with 50% overlap in Matlab does
not pass the constant-overlap add (COLA) test.
However, based on a well-referenced document on this topic Heinzel et al.
(2002), Hanning is amplitude flat at 50% overlap (see p.54).
https://urldefense.com/v3/__https://holometer.fnal.gov/GH_FFT.pdf__;!!Mih3wA!C2vdBEZ9n0h67zOX6inDjk6kdlMMcKgf9h8n78yAnq2S5TY2qS6vKpW3DbikUOVE9C9uTlOZa_oRIz4sLGbowmIrRgE$ 
Do you know why?

I ended up using Hamming, 50% overlap, with 'ola' to be COLA compatible.

Makoto

On Fri, Jul 28, 2023 at 8:30 PM Franziska Kirsch via eeglablist <
eeglablist at sccn.ucsd.edu> wrote:

> Dear List Members,
>
> currently I am trying to inverse transform data in time-frequency space
> back to time domain. I have used the istft - function (inverse short time
> Fourier transform) from Matlab, but it does not seem to work. I keep
> getting error messages about problems with the input data, even though it
> should be correct (as the function says ‚matrix with time increasing across
> the columns and frequency increasing down the rows‘). Also, I'm not sure if
> the input requires complex data or data already transformed to power.
>
> Another idea is to calculate it manually and multiply a complex sine wave
> by the complex Fourier coefficient and sum the results.
>
> Any recommendations on how to achieve an inverse transform? Thank you in
> advance for your help!
>
> Yours sincerely,
> Franzi Kirsch
>
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu