“Finale or Sibelius?” is a question that composers love to ask other composers. It’s often taken as a given that if you write music professionally, you’re already using one of these popular notation software packages. This may be about to change—with the news of Sibelius’s development team being unceremoniously dumped by Avid and subsequently scooped up by Steinberg, we may have a third variable to add to that equation. ThinkMusic, another newcomer, promises an iPad app in the near future, but has already generated controversy for seeming to use Sibelius in its video mockup.
In the meantime, there are a variety of other, lesser-known options for notation software already lurking out there. None of them may have the same clout with professionals as Sibelius and Finale—yet—but many are gaining ground. Whether they present robust alternatives for creating notation (MuseScore, LilyPond), or alternative ways of thinking about and deploying notation (Abjad, JMSL, INScore), each has its own advantages and its own dedicated following.
MuseScore: Open Source Upstart
MuseScore started out in 2002 as a spinoff of MusE, an open source sequencer created by German developer and musician Werner Schweer. Until 2007, however, MuseScore was an obscure piece of software only available on Linux. In 2008, Thomas Bonte and Nicolas Froment began to work on bringing the software to a wider audience. Now, over 5000 people download MuseScore every day. Bonte credits the software’s newfound success to its extremely dedicated developers and early adopters. Its open source community now boasts more than 250 contributors adding to the project. This includes making the software available in new languages, fixing bugs, writing documentation, creating video tutorials, and so on.
MuseScore 1.2 downloaded over 1 million times in 8 months! flic.kr/p/dxg33S
— MuseScore (@musescore) November 29, 2012
While Bonte admits that MuseScore is not yet as feature-complete as Sibelius or Finale, he highlights the price tag: MuseScore is completely free, while the others can run as much as $600. Bonte also points out that when compared to the others, MuseScore is a fairly young piece of software. He anticipates that in a few years, “Musescore will have 80% of other notation software’s feature set on board.”
Another long-term advantage is MuseScore’s open source status, says Bonte:
Anyone can look into the code, change it and distribute it further. This is not possible with proprietary software like Sibelius, Finale, and Score. Given the recent uproar in the Sibelius community about Avid closing the London office, it seems now more than ever appropriate to say that choosing free and open source software is the right thing to do. What happened with Sibelius may happen with any other proprietary software, but cannot happen with MuseScore or LilyPond. The source code is available to everyone; no one can take it away.
This openness made MuseScore the notation software of choice for the Open Goldberg Variations, a project to create a new, quality edition of J.S. Bach’s beloved work that would be freely available in the public domain. This time, the venerable work had a very modern path to publication: the project was crowdfunded through Kickstarter and remained open for peer review on musescore.com before being made available for download. The Open Goldberg Variations can be found on the IMSLP / Petrucci Project website, though anyone is welcome to host or share it.
Musescore.com is MuseScore’s latest initiative. Launched in the fall of 2011, musescore.com is an online sheet music sharing platform, and the only thing that MuseScore charges for. Bonte compares the business model of the site to Flickr or SoundCloud—subscribers pay a fee ($49 per year) for more storage and features, essentially. Bonte says this revenue stream allows them to continue to develop MuseScore full time, while maintaining the open source status of the software itself.
LilyPond and Abjad: A Marriage of Composition and Code
Jan Nieuwenhuizen and Han-Wen Nienhuys are the creators of LilyPond, another open source music notation package. The project that would eventually become LilyPond had its genesis in 1992, when Nieuwenhuizen was playing the viola in the Eindhovens Jongeren Ensemble, a youth orchestra conducted by Jan van der Peet. According to Nieuwenhuizen, the players struggled to read from computer printouts so much that they soon switched back to handwritten parts. That got him thinking: “Fully automated music typesetting done right—how hard could that be?”
As it turns out, it was not terribly easy. Using the typesetting system TeX as a foundation, Nieuwenhuizen began working on the problem with Nienhuys, a French horn player in the orchestra and math student at the Eindhoven University of Technology. But it wasn’t until four years later, in 1996, that LilyPond finally emerged after four flawed prototypes. Despite being plagued by difficulties, however, they found that they couldn’t leave the problem alone. “We never realized how hard it was to produce beautifully typeset music automatically until it was too late and we were hooked,” Nieuwenhuizen admits.
Since those humble beginnings, LilyPond has matured into a full-fledged community project, with over 50 authors contributing to the latest stable release for Windows, Mac OS X, and Linux. This includes one full-time developer, David Kastrup, who makes a living—“just barely,” says Nieuwenhuizen—from donations to the project, which Nieuwenhuizen sees as a major milestone.
Because LilyPond is primarily a typesetting and engraving program rather than a compositional tool, its user paradigm differs somewhat from programs like Finale/Sibelius/MuseScore. Similar to Score, the most common engraving program until Finale came along, musical notation is initially entered as text characters, separating out the step of encoding the notation from the act of graphically displaying the notation, while ensuring a consistent layout. Nieuwenhuizen admits that this can be scary or intimidating at first to composers unused to working this way, but contends that in itself, LilyPond is “quite intuitive and easy to use.” He also foresees more community development of graphical front ends, web-based services, and tablet apps that will make LilyPond even more accessible to those just starting out with the software.
This community may be LilyPond’s greatest asset, with a significant amount of overlap between users of the software and those tinkering with the software itself. This new generation of composers who code is extending LilyPond’s functionality into unforeseen territory. For example, Victor Adán, Josiah Oberholtzer, and Trevor Bača are the lead architects of Abjad, which allows composers to write code that acts on notation in LilyPond in “iterative and incremental” ways. In other words, instead of creating notation directly, composers write code that Abjad then translates into a format that LilyPond can interpret to generate notation. As a result, instead of just manipulating individual notes and objects, Abjad can manipulate higher-level structures—like changing the dynamic level of every instance of a particular note, to give one basic example. Abjad uses the Python programming language, known for its readability and flexibility, as its foundation.
Writing music with Abjad presents a departure from the traditional compositional process. For Bača, it occupies a position “somewhere between the notation packages like Finale, Sibelius, and Score, and the composition environments like OpenMusic and PWGL.” He describes the process of working with Abjad as a “two-part loop,” alternating between writing code to model parts of a score and considering the notation as visualized in LilyPond. This iterative process of continual revision blurs the boundaries between programmatic and musical thinking, as well as between composition and pre-composition.
The creators of Abjad have also worked closely with practicing composers in the course of development. One of these, Jeffrey Treviño, is already well versed in the musical uses of technology; in the course of writing Being Pollen, a work for percussion and electronics based on the poetry of Alice Notley, he estimates that he used nine different pieces of software. With Abjad he had a specific application in mind—he hoped it would help him notate the rhythms of Notley reciting her poem. He describes part of the process here:
I used Max/MSP to tap along to her recitation and make a text file of millisecond counts for when each syllable occurred. I tightened these up in Audacity to line up precisely, and then I exported the numbers again. I wanted to use these numbers to make a notation in Abjad, but Abjad didn’t have a quantizer… We ended up looking up some research together, especially Paul Nauert’s writing on Q-Grids quantization, and Josiah ended up making the quantizer for Abjad.
In this case, Treviño’s needs as a composer had a direct impact on the development of Abjad, and this in turn allowed Treviño to accomplish something musical that would have otherwise been impossible, or at least far more difficult. Treviño draws an analogy between this model of collaborative composing and high-level chess:
Remember when it was a big deal that Deep Blue beat [Grandmaster Garry] Kasparov in 1997? No one mentions that they did a tournament after this where people could use computers to assist them. When Kasparov had a computer, he beat Deep Blue—but most intriguingly, an amateur aided by a computer, not Kasparov, won the whole tournament. So, I’m a lot better at writing twenty-part counterpoint that doesn’t break any rules if a computer can help me. But the skill set it takes to get the computer to know the rules is a very different skill set than the skill set we teach students in counterpoint classes. That’s all to say—I think it’s best to think about all this as totally redefining the important skills of the creative act, so that formerly conventional amateur/master relationships might be turned on their heads. Rather than expanding or enabling skills that matter currently, this proposes a totally new set of competencies and approaches to the task.
(N.B.: Your author independently thought of this analogy, so it must be a good one.)
JMSL and INScore: Notation in Motion
Nick Didkovsky, the primary developer of the Java Music Specification Language, is a guitarist, composer, and programmer who leads the avant-rock octet Doctor Nerve and teaches computer music classes at NYU. But for many years Didkovsky’s parallel interests in music and computers remained independent, never intersecting. What finally inspired him to combine them was an article by Douglas Hofstadter in Scientific American about game theory and a particular kind of lottery called the Luring Lottery, in which the collective desire to win is inversely proportional to the amount of the prize. Didkovsky says, “[The Luring Lottery] is a beautiful and simple idea that binds people together in a simultaneously competitive and cooperative relationship… I wanted to realize that structure musically and thought computers might need to be involved.”
He turned to Pauline Oliveros for help, and she directed him to Larry Polansky. Polansky, together with Phil Burk and David Rosenboom, had created the Hierarchical Music Specification Language (HMSL), a programming language offering a suite of musical tools that turned out to be perfect for Didkovsky’s task. Today HMSL might be most easily compared to other audio programming languages like Max/MSP and SuperCollider, but in an era when these languages were in their infancy, what appealed to Didkovsky about HMSL was its open-endedness: “You can basically do anything… no two HMSL pieces sound even remotely the same because you’re not starting on a level high enough to influence musical tastes. It’s a very non-stylistically biased environment for musical experimentation. And so I think it’s kind of deliberate that it’s kind of a tough environment to work in, or at least it just doesn’t come with a lot of bells and whistles.”
For the next ten years, Didkovsky continued to develop music software with HMSL on the Commodore Amiga for Doctor Nerve as well as other ensembles like the Bang on a Can All-Stars and Meridian Arts Ensemble. When the Amiga platform began showing its age, Didkovsky and Burk had the idea to rebuild HMSL in Java, which could be run on multiple platforms, and in 1997 Java Music Specification Language was born.
The most significant change to JMSL since those days is the addition of a music notation package. With his commitment to traditional instruments, it made sense to Didkovsky to use JMSL to drive a notation environment—and the result was, in his words, a “huge catalyst” creatively. In addition to the many pieces Didkovsky has written using JMSL since then, it has also become a tool used by composers all over the world:
One of my former students, Peter McCullough… developed an extensive suite of personal tools that did very unusual things to scored music, designing both generative and mutator routines that populate the score with notes and transform them once they are present… [progressive metal guitarist and record producer] Colin Marston wrote a series of notated pieces that are impossible for a human being to play—superhuman, intensely noisy pieces performed at high tempo that just rip by and throw glorious shrieking noise in your face, while the staff notation is populated by thick clusters of notes flashing by.
Didkovsky is quick to note that, while traditional staff notation is an important feature of JMSL, it represents only part of what the software can do. Many of the applications of JMSL have little to do with standard music notation—for example, the Online Rhythmicon, a software realization of the instrument Leon Theremin built for Henry Cowell, or Didkovsky’s MandelMusic, a sonic realization of the Mandelbrot set.
Nonetheless, JMSL’s notation capabilities may end up being its most widely used feature, especially with the advent of MaxScore. Didkovsky collaborated with composer Georg Hajdu to create MaxScore, which allows JMSL’s scoring package to communicate with the more popular audio programming environment Max/MSP. Currently, most of Didkovsky’s development energies are directed towards improving MaxScore.
INScore, created by Dominique Fober, is a similar synthesis of ideas from notation software and audio programming, though Fober is quick to stress that it is neither a typical score editor nor a programming language. Fober is a musician with a scientific background who found himself doing more and more research related to musical pedagogy. He now works for Grame, a French national center for music creation, where he conducts research related to music notation and representation.
INScore follows from Fober’s experiments based on the idea that, by providing immediate feedback to the performer, musical instruments act as a “mirror” that facilitates learning. Fober wanted to design a musical score that could act as a similar sort of mirror of musical performance, in the form of graphic signals informed by the audio that could augment traditional musical notation. Fober refers to this approach as an “augmented music score.”
“There is a significant gap between interactive music and the static way it is usually notated,” says Fober. Even with live electroacoustic music, performers generally read from paper scores that give an approximation of the electronic events. There are tools like Antescofo that allow computers to follow a score, and tools for the graphical representation of electronic music, like Acousmograph and EAnalysis, but INScore’s approach is different. “[With INScore] the idea was to let the composer freely use any kind of graphic representation—not just symbolic notation but images, text, and video as well—to express his or her thoughts in a form suitable for performance.”
Montreal-based composer Sandeep Bhagwati used INScore for an entire concert of works entitled “Alien Lands” in February 2011. Meanwhile, British composer Richard Hoadley has written Calder’s Violin for violin and computer, premiered in October 2011. Calder’s Violin uses INScore to dynamically generate the violinist’s score in the course of the performance. INScore is not solely aimed at composers, however, and it has also been used for pedagogy, for sound installations, and to model analytic scores of electroacoustic music.
The Future of Notation?
Despite the vast differences in all of these notation software packages, one thing that they have in common is that each offers something, small or large, that Sibelius and Finale don’t. If you’re looking for something easily accessible and free, MuseScore and LilyPond are well worth checking out. If you’re interested in algorithmic or interactive notation and are willing to deal with a somewhat sharper learning curve, Abjad, JMSL, and INScore are capable of remarkable things. Not to mention the many options I haven’t discussed—BACH Automated Composers Helper, Heinrich Taube’s Common Music and FOMUS, IRCAM’s OpenMusic, and the Sibelius Institute in Helsinki’s PWGL. With all of these tools at our disposal, chances are we might not be hearing “Finale or Sibelius?” for much longer.