I would like to try and create my own audio dataset which I can then use to train machine learning models for classification. The data that I've gathered consists of multiple long audio files of around 1 hour each, containing audio of different categories. Since this is my first time working with audio files instead of data in a tabular format, I am a bit lost on how the do the labeling/preparation.

Most information I find on the internet is mainly related to applying machine learning models to existing pre-labeled datasets. I am hoping to find some more information specifically about what would be the best way to approach the labeling and, if possible, also some information regarding the division of the long audio files into much shorter audio snippets.