Import Error from Scipy Library 'kaiser' and asserting sampling_rate of audio wrongly

[004Ajay]

I was correcting some errors for another person, and I'm just a casual coder, not an expert. So pardon if I say something foolish.

I installed the package using pip install parallel-wavegan

Scipy Kaiser Import Error

Error:
ImportError: cannot import name 'kaiser' from 'scipy.signal'

From the file:
.../python3.10/site-packages/parallel_wavegan/layers/pqmf.py

• I tried correcting your code locally, see the err-making code here, Line 11(import statement) and 45(function call)
• Used the Scipy(1.14.1 stable) documentations see here and found three functions for kaiser: kaiser_atten(), kaiser_beta() and kaiserord()
• I changed and saved the function name as kaiser_atten() and kaiserord(), and worked out to get the two outputs for both the functions.
• Since I am not an expert in this, I wasn't able to differentiate the generated outputs of kaiser_atten() and kaiserord()

Sampling_rate Error

Error:
fs == config["sampling_rate"]
'audio.wav' seems to have a different sampling rate.

From the file:
.../python3.10/site-packages/parallel_wavegan/bin/preprocess.py

• I think the error is in checking (fs == config["sampling_rate"])
• So I changed == to != that is (fs != config["sampling_rate"])
• See the err-making code here

---

Ending Notes:

• I used these steps to run your code (from your main README)
• Even though these small steps have avoided the error block for me. I would like the maintainer to check if these changes are valid and align with your logic.