Computer Science

iTunes Meets Wikipedia

See allHide authors and affiliations

Science  11 Aug 2006:
Vol. 313, Issue 5788, pp. 737
DOI: 10.1126/science.313.5788.737c

The organizing efficiency offered by searchable electronic databases has long been among the most useful features of modern computers. Compared with organizing text files, however, assembling a searchable multimedia database of recorded music is a daunting task. A musicologist would like to be able to type in “Mozart” and “piano sonata” and get as output a list of recordings sorted by performing artist and a selection of stored musical scores. The researcher might then like to synchronize each recording with the score so that when replayed, the recording would follow the score precisely as shown on the screen. Ideally, playing a few notes on an interfaced musical keyboard would cause the system to zero in on a particular passage.

Dunn et al. explain that such a fully functional system may be a decade away from realization. Nonetheless, their work on a system called Variations2 is gradually leading to more powerful music storage and retrieval environments, in which nontextual objects such as sound recordings are linked with graphical objects such as musical scores (which may exist in numerous editions) and the underlying sequences of musical notes. The researchers say that the next version, Variations3, will improve content-based searching of musical works and add better support for non-Western music. Such research could also provide valuable general strategies for navigating a wide range of nontextual data. — DV

Commun. ACM 49, 53 (2006).


Navigate This Article