The Guardian (Manchester); 03 May 1990; Claire Neesham; p. 31
PROFESSIONAL music publishers have been slow to accept comput erised typesetting. Most music is still set by music typewriter, by using Notaset (a Letraset-like transfer system), or by cheap labour in Korea. There are only five traditional engravers still chiselling out music on pewter plates. The reasons for this include the small size of the industry and the competitive rates charged by the many freelance manual copyists. But Philip Sparke, chairman of the Music Publishers' Association technology panel, says: ``The main problem holding back computerised music typesetting is the systems have not been perfected yet.'' Music looks as if it ought to be easy to typeset automatically. It is, after all, only a series of blobs and squiggles whose positions are defined mathematically. But placing the blobs and squiggles on a page is more an art than a science. For example, it is easier to read music when notes are not mathematically spaced and each repeat is set slightly differently. Today's software technology cannot cope with such subjective layout. Over the last five years numerous desk-top computer systems have appeared, which claim to produce typeset quality music output, though only a handful can produce scripts that approach the exacting standards of professional music editors. These highly-rated systems include Amadeus, developed by Amadeus Music Software in Germany, Finale from Coda Music Software, Leland Smith's Score for the PC, HB Engraver, and the ScanNote system, distributed by Toppan, a Japanese printing company. Having been designed from the typesetting end, rather than being an enhancement to performance software, they provide the freedom to edit and control the placing of notes, so scripts ``look right''. Simon Mathews of Schott & Co, London, says that the only point of having a sheet of music is to make performing easy. The main criteria he looks for in well typeset music are: a script that is restful to look at, correct and accurate spacing of music in bars, compassionate page turns, nicely shaped slurs and ties that don't interfere with the noteheads, and note stems which are the right length and connected by properly angled beams. Software developers have tried to design packages that can automatically meet some of these criteria. For example Amadeus has utilities which can work out good places to make page breaks by monitoring bar lines. Most typesetting-based packages also make some attempt to align text with notes and allow the automatic extraction of parts or text from the score. But even with this type of facility the operators often have to intervene to adjust the angle of a slur or length of a note stem to make the music readable. These systems are far from being fully automated. As Bill Buxton, composer, consultant and user interface researcher, points out, ``If a system is fully automated it is not enough to have a knowledgeable operator, the software has to know about the exceptions.'' Susan Homewood, music editor and co-author of the MPA's new book The Essentials of Music Copying, believes one of the reasons why it is difficult to get music typeset well by computer is that the developers and users are proficient in electronics but not in engraving. The major publishing houses use only a few computerised engravers. One is Barnes Music Engraving in East Sussex. Simon Mortimer, a founder, agrees there is no substitute for operator experience. But he says, ``the system has to get the setting right the majority of the time. You can't afford to spend time lining up every slur with a mouse.'' Barnes uses the Amadeus typesetting software running on DEC PDP-11s and Atari STs, under Idris, Whitesmith's version of Unix. The machines are linked into a Monotype Lasercomp, which produces camera ready artwork. Mortimer and his co-founders Mark Broad and Mike Mack Smith chose Amadeus because they had experience of the software. They have since spent three years helping to develop the system and are confident that they can deal with any problems. Ideally Mortimer would like to have a music typesetting system which could be ``taught rather than programmed'' so that operators without computer experience can alter the system to meet the needs of a particular job. His ideal system would have a simple input language enabling the user to define the graphics styles they wanted to use. And it would be able to play the typeset tunes. Barnes can already convert Amadeus files into Midi (musical instrument digital interface) data, which has proved useful for checking music transcribed from records and tapes, but the method is not completely integrated. They can also define new graphical symbols, although this is a lengthy process. The flexibility to design new symbols is essential for the setting of contemporary composition, where traditional notation has been abandoned. Tom Endrich, a director of the Composers Desktop Project, based at York University, says ``We need real graphics in order to represent electronic sounds so the essential feature we would be looking for in a typesetting system is user-definable graphics.'' Buxton believes this is still some way off: ``The problems of capturing the nuances of music script, show how pathetically inadequate computers are at replacing traditional tools.''