Arabic Music Genre Audio Dataset – Part 2 (Ray, Shami, and Tarab)

Résumé fourni par la source

This dataset is Part 2 of a two-part Arabic music audio corpus developed for research in music information retrieval and Arabic music classification. It contains 600 WAV audio excerpts organized into three categories: Ray, Shami, and Tarab, with 200 recordings per category. Each audio file has a corresponding symbolic state-sequence TXT file available in Arabic Music Genre Symbolic Sequence Dataset – Part 2 (Ray, Shami, and Tarab) (https://doi.org/10.7910/DVN/ZQRJDB). The corpus was constructed using a structured collection and balancing methodology. Songs were manually collected and organized by Arabic music category, and within-category clustering was used during corpus development to identify internal musical variations associated with characteristics such as rhythmic patterns, instrumentation, regional style, and vocal accent. The corpus was incrementally expanded while considering the representation of these internal groups. The middle region of each recording was retained for the released audio representation after preliminary experiments showed it to provide stronger classification performance than the intro and outro regions.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.