Dataset is a very important role in music information retrieval (MIR), it widely used for several task, such as: music classification, music recognition and music identification. Indonesia is an archipelago country, consisting of 38 provinces, where each of province in Indonesia has its own regional songs. In Indonesian culture, regional song is part of the cultural identity and culture of the local community, also have influence in their respective cultures. The motivation of this study is to facilitate researchers in the field of MIR by presents the methods used in the creation of Indonesian regional song dataset. The IRSD aims to overcome this hurdle by providing feature extraction from 500 tracks of Indonesian regional song, from 10 provinces with total 67 features were extracted.