Development of a music score production system on a pen-based computer by using auxiliary music data

在普通的音乐创作过程中，用户使用乐器通过反复尝试来创作音乐短语。为了在手写音乐查询的基础上检索音乐表演中的相似音乐短语，我们提出了用钢琴卷来表示两个音乐数据，然后根据它们的重叠面积来计算它们之间的相似度。在实验中，使用由三个或更多手写音乐符号组成的查询获得了前三名的检索准确率。其次，为了通过手写查询从数据库中检索手写对象的任意部分，我们提出了一种新的基于圆圈重叠区域的相似性度量，其位置和大小由手写笔划的位置和类别类型决定。我们使用一个包含1425个笔画和30个手写查询的数据库来检验该方法的有效性。结果，前十名的检索准确率达到97%。

In a common music composition process, a user composes a music phrase by a trial and error using a musical instrument. In order to retrieve similar music phrases in the musical performance based on a handwritten music query, we have proposed that the both music data are represented by piano rolls and then the similarity of them are calculated based on the overlap area of them. In our experiments, top-3 retrieval accuracy of 70% or more was obtained by using a query consisting of three or more handwritten musical symbols.Next, to retrieve an arbitrary part of handwritten objects from a database by a handwritten query, we have proposed a new similarity measure based on an overlap area of circles whose positions and sizes are determined by positions and category types of handwritten strokes. We examined the effectiveness of the method using a database of 1425 strokes and 30 handwritten queries. As a result, top-10 retrieval accuracy of 97% was obtained.