Beyond MIDI: The Handbook of Musical Codes

Musical Codes for Today's World

Eleanor Selfridge-Field

    Hundreds of codes for music have been developed. Most are intentionally made invisible to users. Not so happily, documentation about many musical codes is extremely scarce. This scarcity impedes applications. It deprives many potential users from investigating the relative merits of different schemes for data representation. It thwarts discussion of generalized representation systems. Worst of all, it imprisons data sets within the confines of the specific applications for which they were created. MIDI is a conspicuous exception. MIDI is widely used, well documented, and works on almost every personal-computer platform. However, MIDI is, among all the codes presented here, the most limited in terms of the number of aspects of music that it represents. It is also the most remote from the music itself, for although it is capable of generating quite life-like electronic performances, its origin in hardware protocols divorces it from the musical concepts that are integral parts of most other codes. Hence the title Beyond MIDI.

    Musical codes can be used to support several application domains. Among them sound, notation, and analysis are the most common and the ones on which we concentrate. While the information sets needed in all three domains have some common features, each has unique attributes as well.

    The codes selected for inclusion in this Handbook are either in the public domain or are otherwise available for use without restriction. Among them, we have favored those in most widespread use at the present time the first draft was outlined (1992-1994), those demonstrating the greatest potential for future use (through 1996), and/or those providing the best models for emulation (also through 1996).

    To give some taste of the range of purposes and systems that codes may serve, we have also included brief coverage of a few codes designed for special tasks and unusual platforms. Although many codes are specific to one platform, this collection of material relates to all operating systems in widespread use in the mid-1990s--DOS, Windows (3.x, 95 and NT), OS2, Macintosh, NeXTStep, and UNIX--and some platforms that are better known abroad (e.g., Acorn and Atari) than in the U.S.

Uses of Musical Codes

    The emphasis of the Handbook is intended to be on practical concepts. While the Handbook cannot serve as a complete reference for any one code, it is designed to cover the basic features of pitch, duration, articulation, dynamics, timbre, and other defining features of music. It also describes the file organization for each code, existing applications, data archives (where relevant), existing file interchange provisions, published references, and sources of further information.

    Since, apart from the de facto use of Standard MIDI Files, there is no widely accepted method for the interchange of musical information, either within one family of applications or across the applications spectrum, we are pleased to be able to offer an entire section on nascent standards (two of them newly available) for data exchange. The Handbook should be a valuable resource for those who wish to evaluate the issues involved.

    Inevitably, some codes for which readers may seek a description are absent. We cannot describe codes that are proprietary(1). We have not attempted to describe audio formats, which are covered in numerous other sources (e.g., the WAVE format) . We have not attempted to cover languages for sound synthesis, which have been well served for 20 years by the Computer Music Journal. Codes which were seminal in their time but are no longer in common use, such as Music V and MUSTRAN, are described briefly in the glossary.

Users of Musical Codes

    Apart from supporting musical applications, the present collection of musical codes may offer useful insights to those interested in text/speech applications requiring coordination of the sound and graphics domains. Such aspects of speech as pitch, accentuation, pace, and inflection all have parallels in music; to date "talking text" software is not highly evolved and texts that speak electronically in local dialects or selected voices (capabilities for which electronic performances of music must provide the analogues) are unknown. This collection may also be of interest to those concerned with a broad range of other theoretical and practical questions outside the immediate confines of music applications. At its base, however, the Handbook is intended to be accessible to ordinary musicians. Wherever possible it takes musical (rather than technical) definitions of musical practice and theory as its primary bases.

    Beyond MIDI originated as an initiative of the Study Group on Musical Data and Computer Applications of the International Musicological Society, which has been co-chaired since 1987 by Walter B. Hewlett and myself. Many of the society's members are interested in the possibility of searching large databases of musical information, in producing new editions from raw materials in machine-readable form, in using online scores for classroom teaching and assignments, in verification of theoretical models, in simulation of performance practice and of compositional techniques, in diverse methods of musical analysis, and simply in easy management of and online access to sources that traditionally have consumed miles of physical shelf space.

    The single greatest impediments to such activities are the general absence of such databases and lack of access to those that exist. The original goal of the group was to facilitate translation from one code to another so that such resources as already existed could be pooled. The contents of this book demonstrate why this is not an easily achieved goal. We continue to believe, however, that it is possible.

CCARH and Music Representation

    The interest of the Center for Computer Assisted Research in the Humanities (CCARH) in musical codes was stimulated in the early 1980s, when we were attempting to design a robust system for creating virtual databases of classical music. These encoded databases, collectively known as MuseData, now have their own Website (in-progress) at Similarly, our interest in systems to represent notated music was peaked by the ten years we spent soliciting annual contributions from 80 or so notation software developers for the music-printing "gallery" of Computing in Musicology [see]. Some selected specimens are viewable online at Since 1994 we have been teaching courses at Stanford University on musical information and music representation. Descriptions are available at and


1. The most notable omission is the proprietary Enigma Transportable File Format used with the commercial scoring program Finale; permission to describe this code was not forthcoming from Finale's owner, Coda Music Software.