NEW INITIATIVE

CompMusic is a European research project that studies several world art music traditions in the context of information technologies. The project started last year with a focus on the music of India (Hindustani and Carnatic) and Turkey and will soon include the art traditions of China and of the Maghreb. It aims to develop information technologies to automatically analyse and describe music, and to facilitate the cataloguing and discovery of music recordings within large repositories. CompMusic is coordinated by Prof. Xavier Serra from the Universitat Pompeu Fabra in Barcelona, Spain.

It includes two Indian research teams, one from IIT-Bombay led by Prof. Preeti Rao working on Hindustani music, and another from IIT-Madras led by Prof. Hema Murthy working on Carnatic music. The field of Music InformationTechnologies has developed rapidly in the last twodecades, following the developmentof the Internet and the growth of theonline distribution and consumptionof music. This has opened up newopportunities for accessing andlistening to music, and musicIT in the study of music traditionsinformation technologies aresupporting new ways to managelarge collections of music and musicrelated data. Their practical use includes the automatic organisationand cataloguing of audio recordings,the efficient search of music datawithin large repositories, and therecommendation of new songsto music lovers. Given that thesetechnologies have to “know” aboutmusic, the research and developmenthas to be done by engineers withthe involvement of musicians andmusicologists. At present, theavailable technologies and mostresearch efforts have focused onWestern commercial music andpractically nothing has been doneon such art music traditions asHindustani and Carnatic music.This is the gap that the CompMusicproject is addressing.In order to approach the study of agiven musical repertoire from theinformation technology point of view,there is a need to have well-structureddigital data available to be processed.In the context of CompMusic, a largecollection of audio CDs from thedifferent cultures is being collectedand open on-line repositories are beingused to gather other types of data.For example, MusicBrainz.org is usedfor the metadata of CDs, Wikipedia.org for descriptive information,and Freesound.org for the specialrecordings that are made. Anyone cancontribute to these repositories andthe contributed information is openand available to everyone, except forthe collected commercial CDs thatcannot be distributed.The second CompMusic Workshopwas conducted on 12 and 13 July atIstanbul in Turkey.

The workshopin Istanbul was the first event thatbrought together all the researchersand music collaborators ofCompMusic. It also invitedresearchers working on the sametopics – a total of around 40 peoplefrom a few countries. It included28 talks and several meetings priorto the workshop to discuss specificresearch and musical topics in aninformal and focussed fashion. Oneof the fundamental goals of theworkshop was to bring togetherengineers and musicians working inIndian and Turkish music to discussthe future direction that the projectshould focus on. Apart from theresearchers of the project, severalTurkish musicians and some wellrecognisedIndian musicologists andmusicians, like Dr. Suvarnalata Rao,T.M. Krishna, and M. Subramanianparticipated.In the case of Indian music, thetechnical topics presented anddiscussed mainly related to tonicdetection, intonation characterisation,melody and rhythm description,community characterisation, andmusic browsing and discoveringmethods.The tonic in a concert is the referencepitch from which all the harmonicand melodic relationships areestablished during a performance.Its automatic detection is anecessary step for analysing themusic, and the results obtained andreported at the workshop showedthat the computational solutionsin hand could be used in practicalapplications. Another musicalelement, the intonation of aperformer in a piece, is not such asimple concept and its automaticcharacterisation is not yet a solved problem, but the initial resultsreported at the workshop were quitepromising.

There were many talks relating to the characterisation of melodies. Raga music has particular ways in which melodies are constructed, very different from other music traditions, and its characterisation requires specific approximations. There is a long way to go in developing proper parameters with which to identify musically meaningful melodic elements, but the research presented in the workshop reported some initial results and good approaches to be further developed. The rhythm in Turkish music revolves around the concept of ‘usul’ and Indian rhythms are based on the concept of tala. These concepts have been formalised quite well by musicologists, but there is practically no computational work on them. Two of the talks focused on some preliminary computational work, but more research is needed before the existing musicological formalisation can be turned into algorithmic approaches to describe rhythm. Within CompMusic, research work is on to characterise and model the communities that support a given music – their taste, opinions, and in general their particular views on music. Computationally this can be approached by analysing on-line communities and by developingmethodologies to extract musically relevant knowledge from them.

One of the presentations was a study on how using the rasikas.org online forum, some musically relevant information was automatically extracted. A highlight of the workshop presented and discussed extensively during the meetings pertained to the developed tools and methodologies that can be used in practical music applications. The interest was on applications for browsing audio music collections in a way that could be of interest to music lovers, musicians and musicologists. One of the talks focused on the initial web-based application being developed by the CompMusic consortium with which a user will be able to play music while discovering patterns and musical relationships, both in the songs and between the songs of the various music collections.

The workshop proved to be a success. It was a great learning experience for all the participants. The exchanges between Turkish and Indian researchers and musicians was fruitful and the participants left Istanbul with the impression that CompMusic would soon make an impact on the future of their music.

XAVIER SERRA

(Associate Professor of the Department of Information and Communication Technologies and Director of the Music Technology Group at the Universitat Pompeu Fabra in Barcelona)

Author

Top