audio processing block

Hello, community.

Suppose I wanted to do some custom audio processing. Is it possible to create a custom mathematical function/algorithm that works on raw audio to produce some other numbers as output? The scenario could be for example an audio classification, but where I want to use custom feature engineering. This would entail, for example, inserting a custom processing block between the raw data and the actual machine learning block. Just to set the background, I'm okay with general coding in C, say, but I'm new to app inventor or coding android (or any other Java) apps in general. Any concrete examples would be appreciated.

Thanks, Ricky