Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Oriented principal components analysis for the fingerprinting of music scores

 
: Rodríguez Arrios, F.J.
: Schmucker, M.

Navarra, 2006, 88 S.
Navarra, Univ., Dipl.-Arb., 2006
Englisch
Diplomarbeit
Fraunhofer IGD ()
fingerprinting; music score; automatic identification

Abstract
The main objective of this diploma thesis is the analysis of a fingerprinting system for the automatic identification of music scores based on oriented principal components analysis (OPCA) and its comparison with the existing fingerprinting system based on principal components analysis (PCA). The selection of suitable features to be extracted from the music scores was made for the previous fingerprinting system and it is enhanced with new additional features. The OPCA is used as a dimensionality reduction method for the features. The compression capacity of the OPCA is tested and compared with that of the PCA, searching for a compromise between robustness and data compression.
Compactness of data is a requirement to be met because the results are supposed to be stored in databases. However, the robustness of the identification method must be kept and a compromise between these two parameters must be found. Moreover, the identification of distorted music scores is tested for new kinds of distortions.
At the end, a comparison of the outcomes of both fingerprinting systems will lead to the pertinent conclusions and outlook.

: http://publica.fraunhofer.de/dokumente/N-45262.html