MEANINGFUL MUSIC RETRIEVAL

Given the present paradigm of Music Information Retrieval (MIR) research, one could claim that we will have attained all research goals as soon as we can answer any musical query with 100% precision and 100% recall. But is this really what drives MIR research and what its use consists in? I do not think so. MIR researchers are generally driven by a notable curiosity about music (see for example [6]). And from a user’s perspective, MIR is not so much about extracting a complete set of items of musical information as about experiencing music that is interesting, enjoyable and, to a certain extent, novel or surprising. The motivation for MIR is thus primarily a musical one. Successful MIR research and applications should therefore contribute to our understanding and experience of music. The principal reason for people to compose, listen to, study, play and buy music is that it is meaningful to them. So my message about the future of MIR is to explore and exploit musical meaning, and to create applications that are able to support the human process of musical meaning generation. Even more so, in my vision, human musical behaviour should become one of the foundations of the discipline.