Community Research and Development Information Service - CORDIS

Novel computational tools to better analyse music rhythm

There are certain types of music and rhythms that cannot be analysed using conventional methods. An EU initiative introduced computational tools to better analyse a wide variety of music.
Novel computational tools to better analyse music rhythm
Most computational analysis techniques do not consider the key role melody plays with respect to rhythm. What is more, existing methods are limited in scope because they focus on Western music.

With this in mind, the EU-funded RHYTHMUS (A generic framework for analysing temporal structure in music) project set out to propose dynamic Bayesian models in order to analyse rhythmic structure. These computational tools help to deliver key data on tempo, metre type and timing characteristics of musical styles that is not possible with current automatic analysis methods.

To ensure widespread impact for the models, project partners analysed non-conventional Greek, Indian and Turkish music forms. The models were assessed on both Western and non-Western music to achieve a wider reach.

The RHYTHMUS team developed an integrated Bayesian analysis framework that considers all levels in the hierarchy of rhythm and the effect of melody on rhythm. It can adapt to changing signal features. The models are able to determine with great accuracy the type of metre and to trace the arrangement of the metric sequence to an unfamiliar music signal.

RHYTHMUS advanced the state of the art in rhythmic structure analysis, thus contributing to music information retrieval. This emergent research domain, which draws upon signal processing, machine learning, information and music theory, is being exploited by industry and academia alike to categorise, manipulate and even create music.

Related information


Music, rhythm, RHYTHMUS, temporal structure, rhythmic structure
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top