Published in the "Letters" column, Computer Music Journal, v. 7, #2, pp. 5-6, Summer 1983

McLeyvier Corrections and Update

I wish to correct some impressions made by the review of the McLeyvier music system in CMJ, Vol. 6, #2. In part, the instrument may not have been seen in full by the reviewer, and some of the information in the review was incorrect. In part, the instrument has undergone considerable evolution during the past year, such that it is no longer what it may have been when reviewed.

The instrument is not just a "music transcription system" (as described in CMJ), but was designed, and has been used, as a live performance and studio instrument, and as a compositional tool. For realtime performance use, the instrument has a group of user assignable Ferrenstat controllers, pedal and other interfaces, and eight channels of low-frequency 8-bit A-to-D conversion for interface with arbitrary control devices, knobs, synthesizer voltages, etc. It also has pseudo-inputs that can be user "patched" in software, such as random number generators. It is not as entirely keyboard based as was assumed, nor was music transcription ever its sole intended function.

Far from being a "microcomputer" as stated in the review, the McLeyvier has a multiple buss multiple processer architecture using PDP 11/23 and M68000 16-bit CPU's addressing up to 512K of RAM. 10 meg of Winchester disk storage plus floppy disk for backup are standard, not optional as you indicated.

At present, the voices are analog, but digital signal generators are well under development, and are scheduled for release later in 1983. These will be 16 or 32 voice VLSI digital generators with high speed hardware multiplication and array processing, to permit realtime operation on audio signal tables, via user-defined coefficients and transforms, and by various interactions of waveforms. These generator boards will be microprogrammable to allow easy software re-configuration for a variety of modes of synthesis.

While the software was developed in the language C (on a VAX 11/780), the McLeyvier presents the user with an extensible command language which provides a basic list of more than 150 (largely compositional and orchestrational) commands. Any of these commands can be grouped in any order and named by the user to form a new command (which can then be used within additional new commands, etc., etc.). Such "macros" can be invoked by realtime input from keyboards or other devices, can call each other, or can simply be typed when wanted. "Synchros" are macros made time-synchronous with composed music, and provide the ability to make all sub-commands within a given command individually synchronous with specific timepoints in musical compositions. A text editor for command development, modelled on the UNIX "ED," is provided, as is disk access in a realtime context (no glitching the audio or rhythm during disk IO) for all types of information used in the system.

For reasons including ill-informed writers, people have the idea that the McLeyvier attempts to completely automate proper score notation from keyboard playing. Anyone who has thought this problem through will know that the notational language is not sufficiently "cut and dry" for this to be possible. The instrument is designed to be useful in the computer's original role - as a labor saving device - by programming it to take over the laborious parts of music notation (extracting parts from scores, transpositions, neat calligraphy...), but must leave certain types of notational decision-making to the human user. A group of algorithms make their best first guesses as to how to notate performances, but in most cases, to be exactly what's desired, a user with a working knowledge of notation's subtleties will want to make adjustments, via a graphical score editer. To keep written music an expressive medium in itself, the computer must allow for personal style in pitch and rhythm "spelling" and for the addition of information (articulations, lyrics, instructions, etc.) not necessarily part of the composition in RAM. There is no single predictable correct way to notate many things composers do these days, and it would be foolish to pretend there were. Music created using the McLeyvier's compositional software will probably want less ex post facto score editing than will transcribed keyboard improvisations, although various ways of dealing with the metric structure deviations inherent in realtime playing are user selectable.

In addition, display of text and limited graphics plotting commands have been implemented. Musically synchronous or non-sync commands can use them and they can be conditional on musical input, making them useful for music education, performance cues, sync prompting in film scoring, and other purposes. More elaborate graphics, including a 16-color buffer and graphics development software and input devices, are tentative, based on demand. (Current graphics are 512 x 512 monochrome.)

SMPTE time-code read-write, and a modem interface for music, instrumentation, and command transmission are in progress, as is an audio sampling system.

Lastly, the phone number given for Hazelcom was incorrect. It is 416-929-0659 in Toronto and 212-224-6660 in New York. The system is modular, and prices vary according to individual configuration.

Thank you for printing this correction/update. Sincerely,

Laurie Spiegel
Hazelcom Industries

Copyright © 1982 by Laurie Spiegel. All rights reserved.

Laurie Spiegel's writings page | Laurie Spiegel's home page