Music Information Research (MIR) aims at producing Information Retrieval and Data Mining methods that when applied to musical data will allow for the support of creative and consuming users' needs. Accordingly, the most basic requirement of MIR is the need to test its methods on real musical data to ensure their efficiency. Original musical works are protected under Greek Copyright Law. Databases are also offered copyright protection as well as sui generis protection (‘database right'). Accordingly, creators are given exclusive rights over their respective creations while third parties are excluded from unauthorized uses of these creations. A lawful use of a musical work/a database is achieved by either obtaining permission from the copyright owner/database right owner or by relying on exceptions/limitations prescribed by law. Greek legislation provides for certain limitations that warrant uses of a work, otherwise deemed copyright/database right infringing. It is debatable to what extent these limitations can be helpful for academic research purposes focusing on musical databases.
TopMusic Information Research (MIR) is a research field that aims to develop methods and techniques for the purposes of retrieval, knowledge discovery and data mining of musical information. The MIR field is also highly interdisciplinary requiring the contribution from a number of fields, such as musicology, psychology, signal processing, information retrieval, machine learning and human computer interaction to name just a few.
The MIR field faces some unique challenges in relation to information retrieval of different data types. The artistic character of the musical domain which, among others includes ornamentation and improvisation elements introduces variations in the interpretation of music data. Moreover, notional entities of music are not easily defined, while the phenomena of polyphony and masking may disguise low-level musical events. At the listening end, the musical cognition is highly subjective, making salient features variable on the level of intensity, the density of the pattern and even the performing instruments.
Although research in music information began in the late 1960’s (Kassler, 1966), the domain only received increased attention since approximately the year 2000 when advances in audio file compression, data storage, network bandwidth, the development of a ubiquitous mobile processing infrastructure and the adoption of Information Communication Technologies (ICT) allowed for increased attention in MIR. Nowadays, all these parameters are at a level where:
- •
Audio compression algorithms, for equivalent to CD audio quality, achieve more than 8-fold reduction (See Figure 1);
- •
Local storage is at ~0.08 euro per Gibibyte while an annual subscription of approximately 60 USD in one of the numerous cloud file-synchronisation services, allows for virtually unlimited volume of remote storage;
- •
Home internet access bandwidth is at the range of ADSL/VDSL, while 3G / 4G technologies allow an equally fast Internet connection on mobiles;
- •
Mobile devices, such as smartphones & tablets, have CPU, storage and power consumption capabilities as well as adoption rates that make these apparatuses truly ubiquitous;
- •
Music notation software can support the professional (co-) development and communication of musical content in common Western-type musical notation (scores).
Figure 1. Comparison of Volumes produced by various compression types leading to audio quality equivalent to CD audio, for a 3 minute song
Moreover, since 2000, a very large number of academic journals and conferences have come to exist that fostered the development of theories, research methods and applications for MIR. For example, the International Society for Music Information Retrieval conference (http://www.ismir.net/) is one of the most highly regarded academic conferences under the auspices of which the Music Information Retrieval Evaluation eXchange (Downie et al., 2010), also known as MIREX, is run.