Sparse Non-negative Matrix Factor 2-D Deconvolution for Automatic Transcription of Polyphonic Music

Mikkel N. Schmidt, Morten Mørup

AbstractWe present a novel method for automatic transcription of polyphonic music based on a recently published algorithm for non-negative matrix factor 2-D deconvolution. The method works by simultaneously estimating a time-frequency model for an instrument and a pattern corresponding to the notes which are played based on a log-frequency spectrogram of the music.
Keywordsnon-negative matrix factorization, automatic music transcription, music information retrieval
TypeTechnical report
Year2006    Month August
Electronic version(s)[pdf]
BibTeX data [bibtex]
IMM Group(s)Intelligent Signal Processing


Back  ::  IMM Publications