Closed jasonnicholson closed 3 years ago
Hi, thanks for your input. The Stockwell transform indeed looks interesting, however I don't have any experience with it. Due to the varying window length, I think it would be challenging implementing it in a fully vectorized way (as my STFT is implemented).
Regarding the use of STFT to estimate transfer functions - one has to be careful with choosing the right windows. You can see this work for more information. If your system is stationary, or if you can assume some parametric model, consider using other, more appropriate methods for system identification.
Thank you for the response. I will take a look at the paper.
Robert Brown wrote the Generalized Fourier Family Transform (GFT). He has recently posted on Github in python3. It is an O(n log(n)) algorithm. The STFT and Stockwell Transform (ST) are subsets of the GFT. https://github.com/robb-brown/GFT
You can close this question. Thank you!
Thanks, i'll take a look (I'm migrating to python anyway ;))
Thank you for writing your stft. I find it interesting. I have written similar code but not as vectorized as yours is.
Have you looked at the Stockwell Transform (S-Transform)? I am working towards seeing how an S-transform differs from an STFT when looking at spectrogram.
I use STFT to look at control signals. i.e. current, voltage, pressure, etc. I work with Electro Hydraulic systems. Often I will calculate a transfer function from an STFT of input and output by doing the following:
The S-transform looks to handle the time-frequency tradeoff better than the STFT but I have not been able to prove this yet.