Arabic Music Genre Symbolic Sequence Dataset – Part 2 (Ray, Shami, and Tarab)
Year: 2026
Type: dataset
Abstract: This dataset is Part 2 of a two-part symbolic sequence corpus developed for research in music information retrieval, Arabic music classification, machine learning, and multimodal learning. It contains 600 TXT files organized into three Arabic music categories: Ray, Shami, and Tarab, with 200 symbolic sequences per category.Each symbolic TXT file corresponds directly to one WAV audio file available in Arabic Music Genre Audio Dataset – Part 2 (Ray, Shami, and Tarab) (https://doi.org/10.7910/DVN/ZFNSQC), providing a one-to-one pairing between the symbolic and audio modalities. The symbolic representation was generated using a proposed procedure designed to describe the temporal evolution of musical characteristics within each audio recording. Each recording was divided into successive time intervals and analyzed to identify changes in its musical characteristics. These changes were encoded as a sequence of symbolic states, producing a compact textual representation of the musical progression within the analyzed portion of the recording. The resulting state sequences can be used independently for symbolic music classification or paired with the corresponding audio recordings for multimodal learning. A detailed description of the proposed generation procedure and its implementation will be provided in the associated research publication.
Source: Harvard Dataverse
DOI: 10.7910/dvn/zqrjdb
Authors: Soboh, Lama, elkabani, islam
Institution: University of Cincinnati
Language: English
Topic: Music and Audio Processing
Subfield: Signal Processing
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7204469643