Title :
Modeling musicological information as trigrams in a system for simultaneous chord and local key extraction
Author :
Pauwels, Johan ; Martens, Jean-Pierre ; Leman, Marc
Author_Institution :
Digital Speech & Signal Process. Group (ELIS-DSSP), Ghent Univ., Ghent, Belgium
Abstract :
In this paper, we discuss the introduction of a trigram musicological model in a simultaneous chord and local key extraction system. By enlarging the context of the musicological model, we hoped to achieve a higher accuracy that could justify the associated higher complexity and computational load of the search for the optimal solution. Experiments on multiple data sets have demonstrated that the trigram model has indeed a larger predictive power (a lower perplexity). This raised predictive power resulted in an improvement in the key extraction capabilities, but no improvement in chord extraction when compared to a system with a bigram musicological model.
Keywords :
audio signal processing; information retrieval; music; bigram musicological model; computational load; key extraction capability; local key extraction system; multiple data sets; musicological information modeling; optimal solution; predictive power; simultaneous chord extraction system; trigram musicological model; Acoustics; Computational modeling; Context; Context modeling; Data models; Harmonic analysis; Psychoacoustic models; Chord extraction; key extraction; music information retrieval; music signal processing;
Conference_Titel :
Machine Learning for Signal Processing (MLSP), 2011 IEEE International Workshop on
Conference_Location :
Santander
Print_ISBN :
978-1-4577-1621-8
Electronic_ISBN :
1551-2541
DOI :
10.1109/MLSP.2011.6064602