Salsa Dataset: First Salsa Music Knowledge Base

Salsa is a well-known musical genre and part of our cultural identity. Its origins date back to the 30s of the last century and it has grown in popularity since its origins. Ever since, by different artists in different regions of the world have modified the genre, through different visions of what salsa is, experimenting and adding new instruments and technology. Thus, salsa becomes an intrinsically complex and difficult genre in qualitative terms. However, little we know at a computational level, about which are the acoustic characteristics defining this music, to make it different from the rest of musical genres. In this paper we show the results of a process that builds a knowledge base of salsa music freely available to the scientific community. This base gathers acoustic information of over 20.000 salsa songs. We pretend to use this information to model different characteristics of the genre by means of AI techniques. In addition to making information accessible to researchers interested in salsa music, another important contribution of this project is to provide tools to enlarge the knowledge base with the help of the scientific community. To achieve this, we developed a software that extracts the pertinent acoustic information from songs belonging to users and then it includes them into the knowledge base.

Sarria, G., Mora, M., Arce-Lopera, C. A. (2016). Salsa Dataset: First Salsa Music Knowledge Base. Ricercare(5), 63-72

Check out the Original Article