Hi there. First of all, thanks for all your work creating the ZSA descriptors package. I use it endlessly. My question is specifically in regards to the bark~ object.
I was curious what underlying language it is written in and how your methods of FFT analysis might stack up against others, speed wise, particularly against a dedicated implementation in JUCE. I'm using several [zsa.easy_bark~]s in my patch and experiencing some choppiness in the readings and interference.
Is there anything less efficient about the descriptors implementation because of it being brought into Max?
If I need several audio streams analyzed simultaneously do you think it would be advantageous to conduct them all in one dedicated Max external (or a purely JUCE project for that matter), rather than using multiple copies of the descriptors object? Could they maybe all be multi-processed or something along those lines by JUCE when all executed together?
Any advice on a strategy you might have here would be exceptionally appreciated!