Symbolic Music and Artificial Intelligence

The presence of similarities in music as well as in natural laguage processing, suggest that music could be an interesting possible application for Artificial Intelligence. Since the beginning of the last century many experiments have been done in constructing some kind of "automated" music, and the begin of the computer era have given a significant contribution to this interests. But interest in the so-called "computer music" (ie not only Electronic Music or music composed on computer, but every kind of music that can be represented in electronic formats) are not only in the way of music performances and reproductions. Even before the great diffusion of Personal Computers, Winograd have given a "Linguistic Analysis" of Tonal Harmony for Computers, indicating a grammar approach to music that it is one of the most typical and abstract approach to music. This approach introduced however some problems, concerning an intrinsic well-defined but a priori definition of musical elements and categories. The improvement in Artificial Intelligence techinque suggest instead to abandon a static ”grammarl”
definitions and concentrate on representational aspects. These aspects will not be then postulated a priori, but acquired by experience, that is from living examples, of class membership. Such an inductive approach has benefits on a methodological ground and also supports a variety of useful research lines, looking at the aggregation of existing repositories and the analysis of virtual communities (e.g. customers, buyers, of similar music material), explaning models for technical and social aspects of genre development and definition.

In this scenario Musical Genres are an interesting case of study, because of the intrinsic hard difficulties to systematically describe it and for no complete agreement exists in their definition and assessment.

People

Working Area

• Music Genre Identification/Recognition/Classification
• Developing a Software Platform for Learning Models of Musical Genre (work in progress)
• Representation Methods and Models for Musical Information
• Symbolic Music Representation (MPEG7)

Publications

Classification of musical genre: a machine learning approach, R. Basili, A. Serafini, A. Stellato, ISMIR 2004 International Symposium on Music Information Retrieval.
All the ISMIR04 accepted papers

Related Internal Projects

• Developing a Genre Learning Platform for Musical Retrival
• Presto Space
• Learning Syntactic Semantic Patterns

References

• Franco Fabbri
  • "Browsing Music Spaces: Categories And The Musical Mind"
  • "A theory of musical genres: two applications"
• Documenti MIT (Disponibile solo copia cartacea su richiesta)
• Complete Collection of All ISMIR Papers

Related Projects

Corpora On The Web

• RWC Music Database
• Music similarity information (Columbia University)

Tutorials, Guides, Formats

• MIDI Tutorial
• Music Notation - formats

Events

• ISMIR Conferences

Music Spaces, Communities

Inter-Annotators Agreement

Algoritms & Classification Tools

• WEKA
• Dictionary of Algorithms and Data Structures

Music Theory

• jMusic Education

Web Resources

• C-BRAHMS
• Digital Music Network
• Incomplete List of Resources on User Aspects in MIR
• MIR Systems Home