Dr Aline Honingh (researcher at the Institute for Logic, Language and Computation, UvA)
25 June, 15:30-17:00
Universiteitstheater, Nieuwe Doelenstraat 16, zaal 3.01
"I just love this song, do you know other songs like this?" "What do you mean? Songs with a similar rhythm, or being similar otherwise?"
When is a piece of music similar to another piece of music? And do people agree on this? These are important questions in research on music similarity. Music similarity is a popular topic in the field of computational musicology, but still not well understood.
Since music can be studied by looking at different aspects like melody, rhythm and timbre, music similarity might be studied by studying so-called `sub-similarities’ like rhythm similarity and timbre similarity. In this presentation we focus on rhythm and timbre similarity in the context of Electronic Dance Music (EDM). Since rhythm and timbre are the two most important musical dimensions in EDM, one can wonder whether, by modeling rhythm and timbre similarity, the concept of general music similarity can be modeled from this as well. In addition to these models, we will look at an experiment in which similarity scores were given by listeners. With these scores, the models can be evaluated, and we can learn more about how people rate (rhythm/timbre) similarity.
In a collaboration with company Elephantcandy, an app based on timbre and rhythm similarity was developed, and will be presented at the colloquium.