Credit to Seid Miad Zandavi (s.zandavi@unsw.edu.au) for development of this method. The Python code has simply been adapted from his Matlab implementation.
The package can be installed with pip install psdPy. It can then be used as follows:
from psdPy import psd
import numpy as np
# Replace X with your data
# X should be in the form (samples x features)
X = np.random.normal(size=(10,10))
psd(X)