Why revisable formats matter for our digital scores
Posted by Jim DeLaHunt on 13 Aug 2015 at 10:23 am | Tagged as: culture, Keyboard Philharmonic, music
The Keyboard Philharmonic overview mentions “revisable, symbolic digital notation formats” for music scores. Sometimes people ask if projects like the Internet Music Score Library/Petrucci Project aren’t already putting music scores into “digital” form. Perhaps a clarification will help. Yes, the IMSLP and many other worthy projects are offering music scores in “digital” formats — but not in “revisable digital notation” formats. The difference matters.
The traditional music score is a book. It is notation, printed in ink on pages bound together, and intended to be read by human eyes. The human brain interprets the marks on the pages as bars, notes, rests, sharps, and flats: as symbols, for pitch and duration and loudness, and the other rich information in a score.
Projects like IMSLP — and CPDL, and more and more libraries — are offering scanned images of the pages of the score books, usually packaged as PDF files. A human brain that views the images can still interpret them as symbols for pitch and duration and so on. So yes, these image files are “digital”. However, a piece of software that reads the file will encounter a description of the image as light and dark dots: e.g. “white dot”, “light grey dot”, “black dot”, “dark grey dot”, etc. In order to recognise the score as notes and rests, as symbols for pitch and duration, the software must also go through a visual interpretation and recognition step. This step is difficult and error-prone. One can’t generally be confident that the result is accurate.
There are digital notation formats which directly represent the bars, notes, rests, sharps, flats, and the other rich information in the score. A piece of software that reads a digital notation file will see a symbolic description of the score: e.g. “A measure, consisting of a quarter-note F, and an eighth-note G”, etc. By design, software can read this description reliably. There is no need for a visual interpretation and recognition step. It is this symbolic description which is so valuable for the Keyboard Philharmonic music transcription project.
It is challenging to communicate, concisely, that we want our digital scores to contain a symbolic description of the score, rather than a visual representation. Mathematicians and linguists might understand a term like “symbolic description”. Software engineers might understand the simile that we are making “source code” to the “object code” of the score’s formatted visual representation. These terms aren’t likely to resonate with the general public. They may well understand the relationship between word processor files, which one can edit or revise or copy/paste; and printed documents, which are more or less fixed and visual. That is the distinction which matters to the Keyboard Philharmonic project. Thus, we use the term “revisable format”, or “revisable, shareable digital score” to convey the idea of the symbolic nature of the digital notation format.
So, if a scanned image packaged as a PDF file will not do for digital music scores, what revisable formats are there? Two good candidates are MusicXML and the Music Encoding Initiative. Both are packaged as XML files. They are best edited with music notation software, but are modifiable with a text editor in a pinch. They aren’t perfect formats for digital scores, but they allow us to get started, and they will improve.
Revisable formats have several advantage over scanned image formats, for representing musical scores.
- The revisable format is easier to interpret for software. By design, the format provides the symbolic description of the score in a straightforward way. There is no need for a visual interpretation and recognition step, that scanned image formats would require. Since that step is difficult, time-consuming, and error-prone, being able to skip it makes the difference between software having access to the symbolic description, and not having access.
- As the number of revisable format digital scores rises, it forms a corpus of music scores accessible to software tools. This opens the door to all kinds of software-based services and tools for music scores, analogous to what we do with text. There could be search engines for melodies, tools to compare scores and highlight differences, tools to analyse the musical structure of a score. This will be a rich and exciting arena for innovation.
- New tools for musicians will lead to new ways for musicians to learn and collaborate. Just having a reliable way to compare editions of a score, resulting in a concise description of the differences, will make it easier to find errors in scores, and allow musicians to share performance suggestions in new ways. Eventually this could lead to collaboration hubs for musicians, much like GitHub is a collaboration community for software developers.
- The revisable format allows reengraving the score, that is, laying it out again. The reengraved score can be displayed on screen. It can also be stored as a printable PDF file, but one where the lines and notes are represented as lines and icons rather than as scanned images.
- Reengraving or reformatting the score makes it easier to use on smart phones and tablet computers. This is because we can change the score’s page dimensions. Electronic devices, and computer printers, have different sizes than the paper on which the original scores were printed. Reengraving the score for the actual display size is more legible than a score blindly shrunk to fit. Also, users rotate their hand-held devices from portrait to landscape orientation. Revisable scores can be reformatted (on the fly) to accommodate.
- The revisable format likely represents most scores in a smaller file than do scanned images. A single page of a scanned music score can consume 0.1-1.0 MB. The revisable format file is likely much smaller. So, for that matter, is the reengraved score when packaged as a graphical PDF file.
- The reengraved score will likely be cleaner and easier to read than the scanned image. The revisable format does not contain the smudges and scanning artifacts of the scanned images. Of course, it will be a challenge is to build reengraving software which matches the score’s original engraving for legibility and beauty, because those 19th century hand engravers were skilled!
So, when we talk about digital scores, the term “revisable” carries a lot of importance. It is one of the pillars of the Keyboard Philharmonic’s value proposition.