Are you using this dataset for a or a specific academic challenge ? I can help you with the code to load the files or structure your formal write-up. Language-Based Audio Retrieval - DCASE

Mention the diversity of the audio (natural sounds, urban environments, etc.) and the linguistic variety of the captions.

Explain that the goal is "Automated Audio Captioning" (AAC)—predicting a textual description from an audio signal.

Five unique human-annotated descriptions for every audio clip.