Robust and Adaptive OMR System Including Fuzzy Modeling, Fusion of Musical Rules, and Possible Error Detection

Florence Rossant,Isabelle Bloch

doi:10.1155/2007/81541

Abstract

This paper describes a system for optical music recognition (OMR) in case of monophonic typeset scores. After clarifying the difficulties specific to this domain, we propose appropriate solutions at both image analysis level and high-level interpretation. Thus, a recognition and segmentation method is designed, that allows dealing with common printing defects and numerous symbol interconnections. Then, musical rules are modeled and integrated, in order to make a consistent decision. This high-level interpretation step relies on the fuzzy sets and possibility framework, since it allows dealing with symbol variability, flexibility, and imprecision of music rules, and merging all these heterogeneous pieces of information. Other innovative features are the indication of potential errors and the possibility of applying learning procedures, in order to gain in robustness. Experiments conducted on a large data base show that the proposed method constitutes an interesting contribution to OMR.

Highlights

This paper proposes improvements and extensions to our earlier work on optical music recognition (OMR) [1]
The motivation for OMR is manifold, and possible applications cover several topics addressed in this special issue, including automatic transcription, editing, transposition and arrangement, semantic analysis, fingerprinting, feature extraction, indexing and mining, which are important components of query systems, and can all benefit from symbolic representations
We have described a complete optical music recognition system

Summary

Introduction

This paper proposes improvements and extensions to our earlier work on optical music recognition (OMR) [1]. OMR aims at automatically reading scanned scores in order to convert them into an electronic format, such as an MIDI file, or an audio waveform. This conversion requires a symbolic representation of the score content, achieved through recognition of its individual components and their structure. The literature acknowledges active research in the 1970’s and 1980’s, see, for example, the reviews in [2, 3], until the first commercial products in the early 1990’s The success of these works relies heavily on available knowledge (as opposed to other document analysis problems): reasonable number of symbols, strict location of the staff lines, strong rules of music writing. The problem remains difficult and solutions are generally computationally expensive, even in cases of typeset music

Objectives

Results

Conclusion