Probabilistic Grammars for Music
Rens Bod · UvA-DARE (University of Amsterdam) · 2001
We investigate whether probabilistic parsing techniques from Natural Language Processing (NLP) can be used for musical parsing. As in NLP, the main problem in music is ambiguity: several different structures may be compatible with a musical sequence while a listener typically hears only one structure. Our best probabilistic parser can correctly predict 85.9% of the phrases for a test set of 1,000 folksongs from the Essen Folksong Collection.