Subset of MICM: Persian Dastgah Music Audio Dataset

This dataset contains 7,134 audio samples of Persian traditional Dastgah music, segmented into 8-second mono WAV files and annotated for machine learning tasks. Each track is labeled with one of six major Dastgah subclasses: Shour, Homayoun, Mahour, Segah, Chahargah, and Rastpanjgah. The dataset was curated from the Maryam Iranian Classical Music Dataset (MICM), with Avaz genre tracks removed. All audio files have been resampled to 32,000 Hz and standardized for use with generative audio models such as Audiocraft. Each sample includes metadata with genre, subclass, instrument labels (Tar, Setar, Santur, Kamancheh, Ney, Tombak, Oud, Violin, Daf), and descriptive text to support supervised and generative machine learning applications. This dataset was used for research in automatic music generation and classification for the Persian Dastgah genre. For more information and listening samples, please see:\url{https://github.com/nimamehranfar/FinetuneAudiocraft}