Why do music scanning programs balk at tremolo notation?

If you are a session player, an orchestral musician or a conductor, you know that most orchestral parts, symphonic scores, opera reductions etc. have yards of tremolo style notation. Whenever you get repeated notes, alternating two-tone patterns and the like they are seldom written out longhand.

But try and put this kind of music through an OMR (Optical Music Recognition) package and what do you get? Confusion: the program just can’t recognise it. Puzzling isn’t it. If this kind of notation is so common (and it is) why haven’t the makers of these products got around to incorporating it into their offerings? Well actually the reason is simple – difficulty. An OMR program has to decipher some pretty tiny and complex notation. Think cross-staff beams with grace notes, dots, accidentals, accents. Add to this those little stunted part-beams and you have a pretty busy piece of graphics, and often it can be so cramped and pressed together that its hard enough even for a human to read. Now add to this the possibility that some of those stems may have tremolo bars running through them and the picture becomes too complicated even for the best. Another reason is that tremolo notation is so varied and comes in so many forms – just think of all the different kinds.

But now, at last there is an OMR engine that actually handle tremolo. Its new and you won’t see it as a retail product just yet, but already its available as a library to developers. Its called ReadScorerLib – just search for it. The OMR is at least as accurate as anything out there and soon you will be able to buy the PC app, followed by other platforms.