ACMUS-MIR: A new annotated data set of Andean Colombian music

The music from the Andean region of Colombia has many particularities in terms of tempo, meter complexity and sesquialteras, tuning systems, scales, timbre, and texture. These particularities make its analysis with state-of-the-art music information retrieval (MIR) methods both interesting and challenging. To support future research efforts in the context of Andean Colombian music, a new public data set was compiled and annotated by expert musicologists in Colombia. Our collection comprises three subsets: (1) The rhythm set with tempo and meter annotations, (2) the instrumental format set including number and types of instruments for each recording, and (3) the scale set reporting the musical scale of each audio segment. The data set was conceived with the goal of supporting the development of MIR techniques better suited for extraction of semantic information in Andean Colombian music. We hope that our data set will also encourage a wide range of studies in the fields of musicology, ethnomusicology, and music information retrieval.