Musical archives and song collections are big data. To analyze and sort such huge collections in music ethnology, on streaming platforms or any kind of archives needs computational tools.
The Computational Phonogram Archiving (COMSAR) standard is using Music Information Retrieval (MIR) and Kohonen self-organizing maps of artificial intelligence to analyze music in terms of melodies, rhythms, timbre, tonal systems, etc. Additionally a neural network leans the music and is able to sort it according to the parameters.
COMSAR is implemented in the ESRA archive of the Institute of Systematic Musicology:
On the right is the list of recordings in the archive. On the left is the neural map in which the recordings are included. The map sorts them in terms of similiarity of a musical parameter (here a timbre-based rhythm or groove theory).
COMSAR is open to everyone and one can also upload his/her own recordings and let COMSAR automatically analyze and sort them: