Can we use Bispectrum function for audio analysis?

[monk1337]

Hi, I am trying to implement Bispectrum function for audio analysis and found your repo. But this function takes a light curve object as input, can we make it more general to utilise for other purposes?

How Bispectrum can be use in audio analysis can be found here:

[matteobachetti]

Hi @monk1337, good point. I drafted the relevant functions here: https://github.com/StingraySoftware/stingray/pull/640. If you could help us test it using your own data, it would be great!

[monk1337]

Hi @matteobachetti Thank you for quick reply and help, However I am still confused how to use any of that function, to give an example my data looks like this:

I have audio segments in numpy representation, such as:

For each segment ( Numpy array ) I am calculating the features, such as :

You can run the minimal working example here:

https://colab.research.google.com/drive/1l_sAJqjfukX-Ux4xNCdjMNBiYPuVmxfz?usp=sharing

How should I calculate the Bispectrum Score (BGS) for this numpy array (audio segment )?

[monk1337]

@matteobachetti any updates?

[matteobachetti]

@monk1337 the use case seems very different from ours. At the moment, it's difficult for me to understand how to adapt our infrastructure to your use case. I would need some time dedicated to this issue, which I cannot afford at the moment. If you have time to investigate this, free to propose improvements of the code, though