Being able to identify writers of handwritten music scores is of particular interest for musicologists. Using this instrument, one is capable of ascertaining where, when, why and for whom a certain music score was written or transcribed; including obtaining information about how a musical composition was spread out in the past. On the basis of the individual handwriting characteristics, one is able to identify all writers. Additional details like the type of paper, the ink and the water mark affect the writer identification as well. The Library of the University of Rostock provided its collection of handwritten music sheets which originate from various areas in Europe and primarly from the 18th century. These music sheets are the basis and source of information for the writer identification project.
In order to be able to assign a handwritten music score to its writer, one assumes that each writer has its own individual handwriting. Consequently, the characteristics of handwritings are determined first of all. Musicologists have classified the notation into 13 feature groups which are: clefs, slant, note stems, note flags, note beams, accidentals, note heads, time signatures, bar lines, note beams offset, rests, writing habits and staves. In these feature groups, there are about 80 handwriting features which represent the writing style of a writer. Each feature group represents a hierarchy of detailed characteristics, which build up the concrete features as well as the base elements of a complex figure. The set of all feature hierarchies is called the Feature Base.
You can find more information about the project and its realization, such as information about the underlying data model, the data mining algorithms and system design in the menu on the left. The writer identification system is available at Analysis. For information about how to use it, have a look at Instructions for use.
The project partner "Fraunhofer Institut" has developed a tool for automatic handwriting analysis of music scores.