At Scoring Notes, we usually cover the latest products and news about music notation software and related technology. Music engraving, as we all know, dates back well before the computer age. Sometimes, though, it’s easy to forget that the computer-aided portion of the history book spans back a good long time, too. In this article, we’ll summarize that history and explore a few key moments that led us to where we are today in the field.
Mainframes and music
Before the personal computer revolution of the late 1970s, computers were so large and so expensive that they could only be found in universities and government departments. But almost as soon as there were computers, people were figuring out how to use them for creating and printing music.
In 1952, the University of Illinois built ILLIAC I, the first computer at a US educational establishment. Weighing two tons, it could hold a massive 5.12 Kilobytes in memory, with storage of 64 Kb. Within a few years, Lejaren Hiller and Leonard Isaacson had programmed the computer to generate music for a string-quartet; and so the Illiac Suite of 1957 has the distinction of being the first example of computer-aided composition.
By the start of the 1960s, Hiller had modified a Musicwriter™ music typewriter, so that it could be controlled by the ILLIAC mainframe with batch instructions.
Several other projects in the 1960s used mainframe computers to print music notation by batch processing printing instructions to 2D plotters. However, these were simply sequencing instructions to the printer, rather than calculating the spacing and position of symbols on the page.
The University of Illinois would continue to be a center for innovation throughout the 1960s and 70s. Their PLATO platform (a graphical system for educational software) was shown to researchers at Xerox PARC, who would adopt and improve many of the concepts into their legendary user interface, eventually licensed to Apple. One software program on PLATO was a notation app called Lime, which went full circle by being ported to the Mac in 1985. A Windows version followed, and the app is still available for both Windows and Mac (though 32-bit).
But there was progress elsewhere, too. Leland Smith was Professor of Music at Stanford University, and he started developing his SCORE program at the university’s artificial intelligence labs in 1967; the first printed output, on a 2D plotter, was produced in 1971. Smith describes his own Handbook of Harmonic Analysis, published in 1979, as ‘the first book on music to be typeset by completely computerized means’ (thus carefully excluding other books whose text or music was produced only partially by computer.)
Code in, music out
The first challenge faced by computer music pioneers was how to represent music notation as data inside the computer. One early project was DARMS (Digital Alternate Representation of Musical Scores), established in 1966 by Stefan Bauer-Mengelberg, a mathematician, lawyer and conductor. DARMS-based notation software continued to be used into the 1990s on Unix workstations, notably the in-house software for A-R Editions.
The Plaine & Easie code (spelled as in Thomas Morley’s reference work of 1597) is a method designed to describe the opening bars of musical works in computer catalogues of library holdings. Created in 1964 by Barry S. Brook, it had the stated aim of describing notation with standard ‘typewriter’ characters. To this day, it is used in online library catalogues, such as the RISM database of music manuscripts.
Stanford University’s Center for Computer-Assisted Research in the Humanities (CCARH), run by Eleanor Selfridge-Field and Walter Hewlett, has been at the forefront of digital representations of music since its founding in 1984, overseeing its own MuseData format, which describes ‘the logical content’ of music scores. An archive of music by Bach, Beethoven, Corelli, Handel, Haydn, Vivaldi and others has been created in MuseData.
David Huron created the Humdrum format for computational music analysis, and hundreds of works by over 50 composers has been encoded and archived.
The CCARH produced the journal Computing in Musicology, which for many years described the state-of-the-art and provided notation samples produced by various software — decades before the Scoring Notes web site first got underway.
‘The great thing about standards is that there are so many to choose from.’ Methods of describing music on computers continued to be created. Charles F. Goldfarb, one of IBM’s ‘fathers of mark-up language’, produced the Standard Music Description Language (SMDL) in the late 80s. GUIDO is yet another music data format, created in the late 1990s by Keith Hamel, Holger Hoos and others, which came with its own engine to render its ASCII data into graphical notation. Keith Hamel, Professor of Music at the University of British Columbia, cut his teeth creating one of the earliest notation apps for the Apple Mac, MusPrint. He followed this with NoteWriter in 1988, which evolved into NoteAbility Pro in 1996, for Steve Job’s NEXTStep computers. He continues to develop the program today for Mac and iPad.
NIFF (Notation Interchange File Format) was a binary file format for storage and transfer of musical notation, and an early attempt at interoperability between software, started in 1994 and completed in autumn 1995. It involved many familiar names from the 1990s: Encore, SCORE, Cakewalk, Mark of the Unicorn, NoteScan, as well as Musitek and Opcode Systems. Lime remains one of the few apps that can import files in NIFF format; the optical music recognition software PhotoScore can export NIFF files.
Coda Music (the erstwhile makers of Finale) were originally involved in the NIFF project, but withdrew. Coda’s successors, MakeMusic, would go on to acquire Michael Good‘s MusicXML format, which is now the dominant interchange format for music. And if you thought that was an end to it all, there are still more formats and standards for music and notation being devised, such as the Music Encoding Initiative (MEI) and MNX.
Of course, getting the music in is only half the battle: you then need algorithms for getting them out, onto the page. Donald Byrd (not to be confused with the jazz trumpeter) is celebrated for his 1984 PhD thesis Music Notation by Computer, which is still essential reading for anyone hoping to write notation software. Byrd also devised his own software, called SMUT, which he used to create the musical illustrations in Douglas Hoffstadter’s seminal 1979 book, Gödel, Escher, Bach: an Eternal Golden Braid. He was the principal designer of Nightingale, an early notation app for the Mac. (The source code for Nightingale is now freely available, though somewhat vintage.)
Redoubtable in the 80s
The 1980s was a pioneering decade for technology in every sphere, and music was no exception.
The MIDI standard connected ground-breaking hardware like Yamaha’s DX7 synthesiser and Roland’s TR-909 drum machine. In the first year of the decade, two researchers at Xerox PARC, Severo M. Ornstein and John T. Maxwell, had created Mockingbird, running on the Xerox Dorado system, which is credited as the first ‘WYSIWYG’ visual editor for notation. It also used a synthesizer keyboard for input and playback.
The technologies from Xerox PARC would end up in the Apple Macintosh, which combined a user-friendly graphical interface together with Adobe’s PostScript language and Canon’s laser printer, bringing the Desktop Publishing revolution to the masses. As with all advances in printing technology, music was not far behind.
(In fact, even before the advent of the Macintosh, David Crawford and Jon Zeeff of Michigan University had written a package called Gregory’s Scribe for the Apple II in 1983, allowing the user to create and print Gregorian plainchant notation.)
The first generation of notation software for the Mac included Mark of the Unicorn’s (MOTU) Professional Composer, Electronic Arts’ Deluxe Music Construction Set, and Sonus’ ConcertWare. This latter app is generally credited as the first app on the Mac to display music on screen and print it. Written by Dr. Chad Mitchell, it was released in 1984. In the same year, apps such as MacMusic, MusPrint, MusicWorks, and Music Character Set had joined the roster.
In 1986, Adobe released Sonata, the first PostScript font of music characters, designed by Cleo Huggins. It was the first digital typeface released by Adobe that had been designed in-house and on-screen, rather than from digitizing existing type. (This initiative became the Adobe Original series, with new typefaces by Adobe typographers such as Robert Slimbach and Carol Twombly.) Sonata had only 174 glyphs, and many common symbols had to be constructed from components, such as ottava clefs from separate clef and numeral symbols.
“Sonata was designed to be used by future notation software. Mercifully, Geoff Brown was in the process of updating his application, Deluxe Music Construction Set. When Geoff caught wind of the font, he decided to include Sonata in his 2.0 revision. Adobe was thrilled at the idea of having an application that used Sonata at its release. For the note shapes and character set, I sent samples out to a lot of musicians and publishers for their feedback. Because Geoff was actively incorporating the font in his app, he gave me the most feedback for its use by a developer. There was a fantastic community of support at the time – people who bridged the traditional print world and people who were embarking on digital music printing.”
Huggins came up with a character assignment map that was memorable, because the character shared either a visual or mnemonic similarity. Metakeys, like Shift and Alt, provided related symbols to the base character. As Huggins wryly noted: “The encoding for Sonata is intended to be easy to use for the lowest common denominator, which is a human being, typing at the keyboard.”
The encoding scheme would become a case study in how decisions reverberate far beyond their intended consequences: 36 years later, Finale, Sibelius and other notation apps still use fonts that are based on Sonata’s character mapping. (Though as extra symbols and additional font files have been added, conformity has not always occurred.) Only this year has Finale version 27 begun a transition to the SMuFL Unicode standard.
Sonata was immediately taken up by several notation apps: DMCS, High Score, Professional Composer v2.0, Mark of the Unicorn, Musikrafters, Personal Composer. The ability to output high-resolution music notation using PostScript music fonts issued in a second wave of new software: HB-Engraver, High Score, Nightingale, and of course, Finale.
Finale v1.0 was released in 1988, bundled with Petrucci, a font designed in-house by Steve Peha. After that, the Music Publishers’ Association approached Coda Music about an improved font, which took flesh as Engraver in 1997. This was produced by font designer Bruce Nelson, under direction from Arnold Broido, the legendary president of the MPA. (A history of Finale’s fonts can be found here.) In turn, this was succeeded by Maestro in 1999, based on Letraset forms.
The mid-80s gave birth to several other computers that proved fruitful platforms for music – all of which began with the letter A: Amiga, Atari, and Archimedes. Amadeus, originally written for the PDP-11 by Kurt Maas, was ported to the Atari ST, and used by Henle Verlag as their first computer engraving tool.
Encore was originally created for the Atari ST by Don Williams for the US company Passport Designs in 1984. Versions for Windows and Mac followed. His work has evolved into a new app, called Overture.
The Archimedes was of course the platform for which the Finn brothers originally wrote Sibelius. Ben Finn admits: “We didn’t look at any other software or fonts, because we didn’t know about them. We’d vaguely heard about various things, but this was before the internet.” The rest is history.
Attack of the clones
Finale v2.0 was released in 1991, with a port for ‘the IBM-PC’. It came with a professional price tag of $749. By comparison, SCORE v2 was $995. An IBM PS/2 to run them on would cost ‘from’ $2,300. At the top end, a 20MHz 386 CPU, with 2MB of RAM, and 115MB hard drive would set you back $11,000. (An operating system was extra.) The second-generation Mac II series, on 16 MHz Motorola 68030 CPUs, ran between $6,700 and $12,000. As the 90s moved on, specs got better and prices got lower, as PC ‘clones’ fought a price war, driving down the cost of generic hardware. Software and hardware standards, such as Unicode and USB, were introduced and widely adopted, and the internet tied it all together.
Fonts: the next generation
As part of the move to Unicode, Adobe reissued Sonata as an OpenType font in 2002, but instead of maintaining the glyphs in their original alphanumeric positions, it moved them to the Unicode ranges for Musical Symbols and Private Use. While this might have been technically ‘the right thing to do’, it made the new font useless as a drop-in replacement for the Type 1 version, and still waits for any software to make use of that encoding.
Cleo Huggins’ original documentation for Sonata includes the sentence: “A complete music notation font would have to include an endless collection of symbols; it should span ancient to current notation.” This would seem a prescient description of SMuFL, the Standard Music Font Layout, now employed by Dorico, MuseScore, Finale, and several other apps.
As one door opens ahead, another closes behind: Adobe have also announced that, from 2021, they will stop supporting Type 1 PostScript fonts in their applications. Apple now describes Type 1 fonts as a ‘legacy format’, which is always a worrying designation.
The mainframes and micros of the 20th century are long since gone; and the software that ran on them are now only a memory. On the Mac and PC platforms, there is a lengthy roll-call of applications that are no longer developed nor maintained: COMUS, ConcertWare, DMCS, Encore, Graphire, Igor Engraver, MagicScore, MusEdit, MusiCAD, Music Write, Music Ease, Musikrafters, NoteWorthy Composer, Professional Composer, Personal Composer, SCORE. Even if you can still acquire them, they probably won’t work on today’s OSes.
After 28 years in the business, Sibelius remains the market leader, with veteran’s choice Finale looking forward to its 34th year. But the last decade has brought us price-perfect MuseScore, and coders’ dream Lilypond; while precocious newcomer Dorico continues to break new ground in only five years.
Beyond that, there are a surprising number of notation apps currently in development (or at least available and still working): Canorus, Crescendo, Cappella, Forte, Guitar Pro, Harmony/Melody Assistant, Lime, Mozart, Mus2, Musink, MusicEase, Nightingale, NoteAbility Pro, Notion, NoteFlight, Overture, ScoreCloud.
Even more notation software is being produced for the iPad and Android ‘tablet’ platforms, with Sibelius and Dorico both recently releasing versions for the iPad.
Music notation software was born into academic confines, with software and standards freely shared between peers. When computers became commonplace in home and office, the commercial market opened up, largely using proprietary formats to guard their market. Now, open-source projects like MuseScore and Lilypond have returned to a collaborative, sharing model. Commercial packages like Dorico and Finale are also embracing cooperation with standards like SMuFL and MusicXML.
There is always new ground for pioneers to tread. As long as there are computers, people will keep using them to create music.