Laboratory for Musical Informatics (LIM), Milan, Italy, 19-20 September 2002.

Reviewed by Denis Baggi
Berganzona, Switzerland

MAX2002, the First International Conference on Musical Application using XML, took place in Milan, Italy, at the seat of the Laboratory for Musical Informatics (LIM) on September 19 and 20, 2002. While open to everybody, it was not a typical conference in the sense that it was dedicated to a very specific topic and it sought to attract contributions that should materialize in actual work for the definition of a new standard for using XML (see www.lim.dsi.unimi.it/max2002).

After the introduction by host Goffredo Haus (see Figure 1), director of LIM and Professor of Computer Science in the Department of Information Sciences (DSI) at the State University of Milan, Denis Baggi explained the two main aspects of this project. First, there is the procedure of the Standards Activity Board (SAB) of the Institute of Electrical and Electronics Engineers (IEEE), the largest professional society of people active in the field (www.ieee.org). A subset dedicated to Computer Science is the Computer Society (CS), comprising about 100,000 members (www.computer.org). In September 2001, the SAB accepted a Project Authorization Request (PAR1599) for the establishment of a standard for the encoding, delivery, and reproduction of music, symbolic and sub-symbolic, via World Wide Web, DVD, and CD-ROM, to obviate the inherent deficiencies of existing de-facto standards. These are either incomplete (MIDI), undeveloped Standard Music Description Language (SMDL), or binary and/or proprietary (AIFF, WAV, MP3). The use of XML should make the encoding of music readable and extensible. The IEEE CS Technical Committee on Computer Generated Music, chaired by Mr. Baggi, initiated this project (computer.org/tab/cgm/tc_cgm.htm).

Second, there exists the endorsement of an “abstract,” which represents the first step for a request for research funding, by the global fund Intelligent Manufacturing Systems (IMS), to which Australia, Canada, the European Union, Japan, Korea, Switzerland, and the USA contribute (www.ims.org). The purpose of MAX2002 was also to determine who wants to contribute to the Standards Project and wishes to ask for financing through its own region.

The conference itself took place the afternoon of 19 September and the next morning, while the afternoon of the 20th was dedicated to a brainstrorming discussion to define more closely the objectives of the whole project. Here follows a brief summary of the contributions.

The first paper was by Giorgio Zoia, Ruo-hua Zhou, and Marco Mattavelli of the Swiss Federal Institute of Technology in Lausanne, Switzerland, “MPEG Audio Coding and XML: Samples, Models, Descriptors.” It described some of the already existing work in the direction of the proposed standard, within the MPEG4 and MPEG7 frameworks, and how this can be described using XML, in particular using the mark-up language BIFS (Binary Format for Scene Description). Examples given in the text of the Proceedings are apparently slightly outdated compared to the version available online.

“ Automatic Marking of Musical Dictations By Applying the Edit Distance Algorithm On a Symbolic Music Representation” was the contribution of France Champagne and Guy Tremblais of the Université du Québec à Montreal, Canada. It was about the XML encoding of a melody and of what a music student notates of it during a musical dictation. The original score is then compared against that written by the student, while an algorithm identifies the differences and computes a grade. The results of the program are fine-tuned by examining those of the dictation instructor.

Jacques Steyn, an independent researcher and consultant from Hatfield, South Africa, contributed “Framework for a Music Markup Language.” Some existing XML-based languages were mentioned: 4ML, FlowML, MusicML, MusiXML, MusicXML, which all focus on a subset of Common Western Notation (CWN), and ChordML. After listing some requirements for a proposed language, such as conformity with XML, simplicity, modularity, universality (beyond CWN), the author examined existing proposals: Hy Time, SMDL, and SMIL. He then distinguished among several features of music, intrinsic (core, e.g., notes; periphery, e.g., rendition of a note) and extrinsic (performance, recording in graphic form or sound), pointing out the complexities and their importance in a mark-up language. Modules of such a language were defined, including frequency, time, organization, texture, effects, performance, control, notation, lyrics, MIDI, synthesis, and more, each one explained in its constituencies.

“ A Proposal For a Regular Grammar To Parse Jazz Chord Notation” was delivered by Mr. Baggi of the Centro CIM della Svizzera Italiana (CIMSI) of the University for Applied Science of Southern Switzerland (SUPSI). The author showed a simple program capable of parsing, identifying, printing, and playing the notes of the symbolic notation for extended chords used in jazz. Such symbols are gathered in a harmonic grid, of fundamental importance in jazz, even though it identifies pieces that may have a totally different sound and style, as the author demonstrated with his saxophone. The proposal is to build an extensible, flexible algorithm for parsing such symbols (note: Figures 1 and 2 of this article in the Proceedings contain errors). The relevance to the standard was apparently not clear.

The last contribution of the day was by Maurizio Longari of LIM, “A Case Study: Prepared-Piano Notation in XML.” A prepared piano is a standard instrument with the addition of various kinds of tools: nails, sticky tape, thin layers of plastic, copper, aluminum, etc. Prepared-piano master Pino Devita gave a demonstration and played a number of works (see Figure 2): Wednesday, a “static” piece (meaning that the piano is prepared in advance and then played normally); Industrial Piano, “dynamic” in the sense that the piano is played directly on the strings, with some performance tools; and Muz’ Breath, another static piece. Note that there exist “mixed” pieces. This demonstration concluded the day with a well-applauded performance.

The session of the second day opened with a lecture by Markus Lepper of the Technische Universität Berlin, Germany, “Modeling Music using XML—Some Basic Considerations.” This was a rather deep paper which, based on a body of pre-existing work, showed that there exists a whole body of algebraic modeling underneath coding languages. The author strongly suggested basing the discussion, and the later standardization, on a semantic model, and to use a language with some degree of abstractness to yield easier comparability between competing solutions than the mere syntax of a set of Document Type Definition (DTD). An architecture was then presented, based on the identification of aspects of an object with multiple values from different domains. Lastly, he suggested incorporating a specially restricted kind of lambda-expression into the coding, so that definitions of default behavior could be contained in the standard, together with the configuration of multiple-track views, data transformations, and other aspect values.

Maurizio Longari of LIM presented “Towards a Symbolic/Time-Based Music Language Based on XML,” co-authored with Mr. Haus. The paper first went through an analysis of existing mark-up languages for Symbolic Music Information (SMI), namely: SMDL/Hy Time, MusicXML, MusiXML, MusiCat & MDL, WEDELMUSIC Format, MNML, MML, MuTaTeD, MusicML, and ChordML. It considered layered SMI at the structural, notational, performance, and audio levels, and their possible representation in XML. It then described the proposed Spine, a structure for an (at least) two-dimensional representation of music that allows the definition of relationships among layers, such as notational, performance, and audio. The paper concluded with mention of related topics and open problems.

Gustavo Frederico, from the University of Ottawa, Canada, authored “Actos: A Peer-to-Peer Application for the Retrieval of Encoded Music.” Actos is an application written in Java that uses XML-based ChordML, written by the author, to perform queries on chord sequences. After a brief introduction, the presentation emphasized examples, on-the-spot, of the software demonstrating its performance in retrieving and identifying chord sequences in a given piece of music, both as annotated chords and as a sequences of degrees with cipher notation.

Michael Good, of Recordare LLC in Los Altos, California, is the guru of his existing XML system for music, which he presented in “MusicXML in Practice: Issues in Translation and Analysis.” MusicXML was briefly introduced by the author together with its design techniques and applications supports, showing how it has become the lingua franca for a number of existing music software programs, including Finale, XEMO, SharpEye, KGuitar, and other related ones, as shown in Figure 1 of his paper in the Proceedings. Then, the issue of converters was dealt with, namely, programs that convert from one music format to another, including NIFF (in spite of its errors), MIDI files, and Finale. It was shown that MusicXML can be used for musical analysis, and this was compared with two other approaches, DOM and Xquery, both with their source. In conclusion, MusicXML was shown to cover such needs. Mr. Good, attentive to the commercial importance of these kinds of standards for music publishers, expressed his preference for starting with a concrete application, instead of from definitions in the abstract, to avoid the pitfalls of SMDL.

Lastly, Perry Roland, of the University of Virginia in Charlottesville, highlighted another facet of the problem with his paper, “The Music Encoding Initiative (MEI),” arguing, in a scholarly opposition to Mr. Good, for an encompassing approach unrelated to a concrete piece of software. The Text Encoding Initiative (TEI) is a parallel to the proposed MEI, which refers to Common Music Notation (CMN). The constraints were listed as: comprehensive, declarative, explicit, interpreted, hierarchical, formal, flexible, extensible. It was explained why the use of XML DTD, a known XML technology, is used instead of schemas. A mix of elements and attributes, “milestone” elements to take care of multiple hierarchies, flexibility, extensibility, and re-utilization of existing standards were proposed, together with the need for further evaluation.

The discussion of the IEEE SA Working Group Meeting on Music Applications of XML took place on the second afternoon and was open to everyone. Mr. Baggi explained again the “bureaucratic” intricacies of both the IEEE Standard procedures and the IMS, together with the opportunities for both creating a truly international standard and for receiving financial support for doing it. At the end of a clarifying discussion, a brainstorming exercise was performed by all members. It came out that the most important feature of the standard ought to be the integration of music at all levels: audio, performance, description, sound, score, and the like. This showed that, in spite of the fact that contributions were sought without any particular guideline or constraint, there seems to reign a consensus of what this particular standard should cover. The participants from almost all world continents were thanked for their contribution in Milan, Italy. A Workshop has been scheduled for late Spring or early Summer of 2003 in Ticino, Switzerland.