I’m an amateur opera and symphonic chorus singer. Most of the classical music and opera I perform is old. Not just pre-iPhone old, but usually well over a hundred years old. These works have outlived even the outrageously long copyright terms imposed on our culture by greedy commercial interests. They are clearly in the public domain; they have returned to the shared culture from which they grew.

But when I want to learn a new work, like Verdi’s opera Macbeth or Mozart’s Requiem, why do I find myself paying $24-$40 for a music score which probably cost $5 to print? Why does the book contain stern warnings not to photocopy the contents, even it is little more than a facsimile of a previous edition, which itself is in public domain?  It is because these music score products still cling to a pre-internet business model, based on selling “molecules” (the physical artifact of the book) for a price based on the value of the “bits” (the information or arrangement of notes we call the musical composition, plus the value of the editing, plus the value of the typesetting), and the costs of distributing and warehousing those molecules.

This shouldn’t be. The music itself — the bits, the abstract genius which is Beethoven’s or Mahler’s, not the later editorial changes, or the molecules on which the bits are printed — is in the public domain, so its cost is zero. Volunteers are willing to scan or transcribe old musical scores for free. So a digital file with a score ought to be accessible for the marginal cost of storage, duplication and delivery.  And in an era of cheap disks and high-speed internet, that marginal cost is zero.

Many classical music and opera scores are indeed available, free for the downloading. Below are links to some useful sites for the classical or opera musician to find them. But there’s more. In the digital world, scores should get better, too: more correct, easier to use, more customised. If a fraction of every chorus and orchestra pitched in to ratchet forward the quality of the free scores for music they perform, we could make a huge difference.

Lessons from other parts of free culture

In many other domains of culture and intellect, business models based on molecules are being transformed into business models based on bits. The marginal cost of copying, storage and delivery of digital information is dropping to zero. Volunteers are willing to create and edit information, and share their results freely. Business are being built on information being free instead of barricaded. The result is transformation. This transformation is also coming to music scores.

Let’s look at some parallels. When I was a child in the late 1960’s, our family had not one but two encyclopedias in the house. Each was a set of some  dozen hard-backed books, edited by subject matter experts, consumed by us readers. Now, in the 21st century, Wikipedia dominates the encyclopedia world. It is freely available, built from freely contributed work of us readers (who are also volunteer editors), thanks to clever social norms. It has over 250 language editions, with tremendously broad coverage and respected quality (despite acknowledged weak spots).  Wikipedia itself is entirely on-line, but entrepreneurs produce various derivative printed editions of Wikipedia. In the travel guide space, WikiTravel offers a fully free travel guide to the whole world, while the Lonely Planet folks are building an online body of grassroots content to supplement their books.

Computer software, like music score publishing, is complex and difficult to do well. But the GNU operating system, the Linux kernel, the Apache Foundation, and many other projects have built up a business model for software development that is based on freely available source code and executable software. Profit-making companies like Red Hat and MySQL have sprung up in this free ecosystem, offering value-added services like support, adaptation, or convenience to the free raw materials.

Cultural works, also, are being set free. Project Gutenburg has for decades organised volunteers to transcribe printed public-domain literature into digital text form, where it may be freely copied and read. They offer more than 30,000 free eBooks, from Shakespeare’s Hamlet to Anna Katharine Green’s The Golden Slipper (which I’ve never heard of either, but it does feature a character named Delahunt). A wonderful film, Nina Paley’s Sita Sings the Blues, is freely available for watching or downloading, but the filmmaker makes money by offering creator-endorsed DVDs and Sita T-shirts.

From this ferment have come some useful terms. “Free/Libre” refers to works which are “free as in freedom”, i.e., you have freedom to use, copy, and build on them. This is different from “free as in free beer”, i.e. you pay nothing to get access to the work. Not everything which is free as in beer is free/libre, and not everything which is free/libre is free as in beer.  Creative Commons licenses are one way for publishers to simply and clearly designate their works as “free/libre” in various respects.  “Open source” is a term used for some computer software which is partially free/libre (and herein lies a rich debate). The term “free culture” is emerging to describe music, literature and culture which is in digital form and free/libre.

Whether an encyclopedia article, a book, a software product, or a film, treating digital content as free/libre unleashes a wonderful dynamic.  Instead of struggling with inconvenient digital locks and paywalls, which prevent you from getting access to what you want, free/libre content is yours for the copying. Instead of sharing being forbidden, it’s encouraged. A thousand volunteer contributors bloom (150,000+ in the case of Wikipedia), because they know the fruits of their labour will be freely available instead fattening a selfish company. And yet there is room to make a business on free content, because some things — attention, customisation, timeliness, authenticity — are of value and hard to copy. (For more on this, see Kevin Kelly’s fantastic essay, Better than Free.)

It is hugely refreshing to escape from the cramped confines of unfree into the sunlit plains of free culture. I find myself going out of my way to find free alternatives, to avoid having to crawl back to that unfree world.  It’s high time the scores of public domain music and opera emerged into the free world.

Free/libre music score repositories

A of web sites and projects which are paving the road towards free/libre music scores. First, let’s look at repositories for scanned music scores. Many people are scanning old music score books, which are no longer in copyright, and uploading the scans to these repositories. Usually, the scans are PDF files containing the scanned images. This means you can view the images on your computer, and print them, but you can’t easily correct or revise them.

  • The Petrucci Music library (http://imslp.org/wiki/) is the largest and best of the scanned music score repositories. Previously known as the Internet Music Score Library Project (IMSLP), it contains over 50,000 scores for over 20,000 works in all genres by over 3,000 composers. They have excellent instructions for how to tell if a score is in the public domain, and good style guidelines to keep their pages consistent and usable. If you want to find a scanned score, start here first. If you want to contribute a well-scanned, clearly public domain or free/libre score, consider this site first. However, the only formats they accept for source-form (editable) scores are for commercial tools, not free tools. Upload your source form score files in Lilypad or MusicXML formats to Mutopia.
  • The Choral Public Domain Library (CPDL, http://www.cpdl.org/) limits itself to vocal music, so instrumental works probably should end up at the Petrucci library. It does accept a wider range of formats, including not just scanned music scores, but retypeset scores in source formats for Lilypond and commercial music typesetting programs. It also accepts links to files hosted away from CPDL. At the moment it hosts over 11,000 scores from over 1600 composers.
  • Many other sites host scanned music scores. For instance, I find some useful scores as part of the digital archives at university library web sites. The Petrucci library’s “Other music score websites” list is a good catalog. If you find a scanned score, with free licensing, available at some other site, consider uploading a copy of the files to the Petrucci library (or link from the CPDL if appropriate). Remember, this is free/libre content. Sharing is a good thing.

Next, let’s consider repositories for re-typeset new editions of old public domain scores. These sites accept source-form (editable) files. There is great value in re-typesetting old editions of public domain scores. The resulting new edition will, if you’ve done a good job, be easier to read and have a smaller file size than a scanned image of an old edition. Also, the source form, or editable form, of the score can be re-used in many interesting ways: for special-purpose editions, for analysis of the music by other software tools, etc. To this end, these sites encourage (in some cases require) that the new editions be licensed freely, e.g. by Creative Commons.

  • The Mutopia Project (http://www.mutopiaproject.org/) is a repository for free/libre music scores in Lilypond format. More on Lilypond below. They standardise on Lilypond because it is a free/libre tool, it produces high-quality scores, and it’s possible to convert from many formats into Lilypond. They currently host over 1000 scores.
  • The Gutenberg Sheet Music Project (http://www.gutenberg.org/wiki/Gutenberg:The_Sheet_Music_Project) is an interesting one to watch. It hosts less than 100 scores now. However, it’s an offshoot of the tremendously successful Gutenburg Project for free ebooks (literature in plain text form), so they know how to run this sort of project. They have an excellent organised approach to content production.
  • Wikifonia (http://www.wikifonia.org/) is a repository for lead sheets of songs. A lead sheet is a simplified music score, perhaps enough to sing at a piano with friends, but not enough to publish a vocal score. They use MusicXML as their standard format. Interestingly, they have an arrangement to pay royalties for music they host. This is probably the best home for re-typeset scores of non-free/libre music.

Free/libre music score tools

How do you create an source-form score from a scanned image of an older edition of the score? You use  one of the free/libre music typsetting tools. These programs let you generate a freshly-formatted rendition of a music score.

  • MusicXML (http://www.recordare.com/xml.html) 2.0 is an XML-based language for music score interchange. It is a file format, not an editor program. Because many editors and tools can read in and write out MusicXML, it is a powerful pivot format.
  • Open Score Format (http://openscoreformat.sourceforge.net/) is a music score interchange format. It starts with MusicXML, adds profiles for how to use MusicXML for various types of scores, a packaging method for including MIDI recordings and HTML descriptions with the score, and a standard for metadata describing the score.  It may turn out to displace MusicXML from the role of pivot format.
  • Lilypond (http://lilypond.org/) is a tremendously capable music typesetter, with uncompromising standards for music expressiveness and typeset beauty.  It takes a text rendition of a music score in an admittedly arcane format, and generates a PDF file with the typeset score in whatever layout you wish.  Think of it as TeX for music, if that helps. It provides powerful ways to build a vocal or orchestral score from individual part scores. Because the tool itself is free/libre, its mechanisms are open to public understanding and correction. It is also free as in beer, so it’s cheap to get started with. Its file format, with the .ly extension, is commonly used as an archive format. Three drawbacks: editing on a source text which is compiled is not as accessible as a What-You-See-Is-What-You-Get (WYSIWYG) interactive approach; the source format is quite arcane, though powerful; and while the tool can produce beautiful opera scores, it takes work to organise the source for opera use.
  • MuseScore (http://musescore.org/) is a cross-platform What-You-See-Is-What-You-Get (WYSIWYG) music editing program. It generates MusicXML output. It is available for Linux, MacOS X, and Windows.
  • Canorus (http://canorus.berlios.de/) is another cross-platform What-You-See-Is-What-You-Get (WYSIWYG) music editing program. It is available for Linux, MacOS X, and Windows. Its current version (March 2010) is 0.7, which they describe as beta-quality and not for product work. It is also over a year old, which means the project may not be advancing fast.
  • Denemo (http://www.denemo.org/) is a music input program for Lilypond format scores. It gives you accelerated ways to generate Lilypond notation from the computer’s keyboard, a MIDI device, or from audio played into a microphone. It is in principle available for Linux, MacOS X, and Windows. However, it is delivered as source code which must be compiled on your computer, opening the door for installation glitches that require a software geek to resolve.
  • Audiveris (https://audiveris.dev.java.net/) is software for music OCR, i.e. examining a scanned image and generating music score source code from it. It is written in Java and generates MusicXML output. It offers the potential of accelerating the production of source-form scores for Mutopia using the scanned images of old editions from the Petrucci Music Library and CPDL as input. I have not yet used Audiveris myself.
  • OpenOMR (http://sourceforge.net/projects/openomr/) is another project music OCR software, but it may be moribund.
  • My purpose is to list free/libre tools. For completeness, I’ll mention some well-know commercial music typesetting products: Sibelius, Finale by MakeMusic, SharpEye music OCR by Visiv, PhotoScore music OCR by Neuratron, PDFtoMusic Pro by Recordare. There are more. But they aren’t free/libre, and at USD200-800 each, they certainly aren’t free as in beer.

The tools for retypesetting scores are important because scores in source format can become better. As successive musicians notice and correct errors,  that the the next musician benefits from a more accurate score. By changing page layout and adding annotations, and creating scores with a particular group’s performance cuts, scores can become easier to use. By changing the note size and substituting lyrics in the correct language or translation, scores can become more customised.

So, if you are a musician, I encourage yourself to check out these respositories and tools.  And I ask you to consider helping the move to a more free culture, by scanning and contributing public domain music scores, and by slowly transcribing the free music you use into source form and contributing that. If a few people from every chorus, orchestra, and cast would pitch in each season to transcribe the public domain scores they use, we could grow the collection of free scores dramatically. We all would benefit.