A score-driven approach to music information retrieval
Journal of the American Society for Information Science and Technology - Music information retrieval
| Hi-index | 0.00 |
We are working on several projects addressingdigitalization and restoration of legacy musical archives;We are dealing with several sources of musicalinformation (audio, notation, performance, video, images,etc). On one hand, these sources are similar to those ofusual multimedia databases. On the other hand, somespecific feature of legacy musical archives call for ad-hocsolutions.In legacy musical databases many relations areimplied by the framework (for instance, recordings ofparts of the same piece), but not made explicit in thecontent. To address this specific feature, we propose aThree-level database schema where the bottom level(called Source Level) describes the source material; themid level (called Logical Level) describes explicit andimplied relationships among bottom-level data. Finally,the top level (called Catalog Level) represents thecommon catalog data attached to pieces (composer,conductor, data of recording).To support this methodology, we need a language todescribe musical content. In this article we survey severalmark-up languages (such as SMDL and XML derivedlanguages) and discuss their features w. r. t. ourproposed layered database schema.