Digitizing Braille Music — How We Do It

My last post presented a list of new digital braille titles that had recently been added to our collection. This post is to make clear that we have a program of digitizing our braille music collection, and to describe how we have been doing it.

It might also be good to underscore why we have undertaken this project:  to preserve our scores (especially the more fragile of them) and to streamline their circulation via downloading from BARD.

There is also the benefit that errors in the original transcription (it does happen) can be corrected and the score re-embossed.

Beginning in 2003, the first system or method used for digitization was software created by a Czech firm, called OBR (optical braille recognition). This is a reasonably priced software that is accessible to a blind user with a braille display. The software supports several standard off-the-shelf scanners.  The only modification or non-standard element is a colored plastic sheet that comes with the software and is placed between the braille page and the light source of the scanner.  The immediate result of the scan is a text file, an alphanumeric rendering or translation of the cell dots.  The software can read both sides of the page in a single scan, in the case of so-called interpoint braille.  So, in 15-30 seconds, one can get two pages of braille in digital form. However, the level of accuracy varies, sometimes considerably. There are many factors behind this, for example, the quality of the braille dots (worn and/or pressed down) and the color (darker, lighter) of the paper.  It is rare that a scan will capture a page with 100% accuracy.  So, we proofread every scanned page cell by cell. There are too many details in music scores so that errors can’t be left to the reader to figure out or discern.

The price of a 100% accurate 30 second scan can be high.  Each page, on average, takes 10-15 minutes to proof.  Nevertheless, this method of digitizing has yielded, since 2003, hundreds of wonderful music scores that can be read, downloaded, and embossed anywhere.  This system is still in use.

A second system for scanning braille music began in 2008, with the purchase of an expensive turnkey (hardware and software) operation made in Russia. Instead of a flatbed scanner, this system uses a digital camera mounted inside a large “box.”  The camera takes two pictures of the page, each with different flash units so that an easily discerned shadow is created that can be read and interpreted by the software.  Only one side of a page scan be scanned at a time. 

This system, too, is imperfect, but significantly better than our first system. It is faster, more flexible, and more accurate.  In the process of discerning, editing or fixing any errors, the system displays three windows on the screen for each scanned page. One, at the top, is the actual photo of the braille, used for reference. Another, below it, is a simbraille version of the photographed braille. That is, the braille dots of a cell are represented by black dots or spots and are not tactile. A third window represents the cells of the page with alphanumeric characters, which can aid the required sighted user. 

Corrections to any errors are made in or from the simulated braille window. With a mouse, cells can easily be moved, changed or otherwise modified to match the original page and therefore create a 100% accurate reproduction of the original page.

A sixty page score can be scanned in an hour and fully proofed in another hour if the original scan is good —and most of them are good.  Of the 2000+ titles that have been digitized to date, the majority were done using this scanner.

A third system was acquired only last year, 2013, from a company in Germany.  Like the Russian system, it uses an actual photo of the braille page as the basis for the software’s digitizing. That photo, in a single window, is where any needed editing is accomplished. This system, too, is fairly fast and convenient, and getting better with new software modifications. It requires a sighted user.  Most importantly, the result of the scanning process is a 100% accurate digital version of the original paper score.  This system has become our workhorse going forward.

We have recently become aware of another—American— system of braille scanning that is in development.  It uses an off-the-shelf scanner, like the OBR.  We hope that it can become another player in the somewhat rarefied field of scanning braille.

In conclusion, the Music Section I inherited a decade ago had only paper or thermoform braille scores, many of which were limited in number, crumbling, and irreplaceable. Now, with a good stock of digitized scores and more constantly on the way, we can replace, correct, and easily circulate newly embossed copies, and make them available for download. This is a very different–and better–world of braille music.