ewtpy

A Python implementation of the empirical wavelet transform for 1D signals, splitting them across N scales through the EWT1D function and a set of boundary helpers.

The package applies the empirical wavelet transform to a one dimensional signal and splits it over N scales. The main entry point is EWT1D, which hands back the transform itself, the filter bank, and the detected boundaries. Around that core sit smaller routines for boundary detection, boundary completion, the Meyer filter bank and Meyer wavelet, the beta function, and the LocalMax and LocalMaxMin detectors.

The README is upfront about what was left out. A few capabilities of J. Gilles' MATLAB toolbox never made it into the Python port, including the transform for 2D inputs, preprocessing, and adaptive or ScaleSpace boundary detection. Anyone coming over from the original toolbox should check those gaps before switching.

Installation follows the old school path. You download the project from GitHub and run python setup.py install from the project folder. The README does not mention any extra dependencies or configuration steps beyond that, so it is about as plain a package install as you can find.

Users who find the package useful are asked to cite it. The recommended reference is a 2020 paper by Vinicius Carvalho and colleagues on adaptive decomposition methods for EEG seizure detection and classification, published in Biomedical Signal Processing and Control. The full citation is spelled out in the README so nobody has to hunt for it.

Questions and suggestions go to vrcarva@ufmg.br. The README also leaves the door open for contributors: build a new feature or fix something and send the files along with the credit line you want included. The author block places the work at the Federal University of Minas Gerais in Belo Horizonte, Brazil, with ties to the neuroscience center.

• Source 1 · Project repository
• Source 2 · README