ieee transactions on audio, speech, and language processing march 2010 lan-ying yeh 99.04.12
TRANSCRIPT
Introduction
Music Information Retrieval (MIR) Singer identification & Vocal-Timbre-
Similarity Feature extraction Influence from other instruments
Related Studies Using a statistically based speaker-
identification method for speech signals in noisy environmentsFirst estimated an accompaniment-only
model from interlude sectionsa vocal-only model by subtract the
accompaniment-only model from the vocal-plus-accompaniment model
Assume singing voice and accompaniment sounds statistically independent Not always satisfied
Estimation have problem
Related Studies
Using vocal separation methodSimilar to their accompaniment sound
reduction method
Did not dealt with interlude sectionsConducted experiments, using only vocal sections
Accompaniment Sound Reduction F0 estimation
PreFEst (Predominant-F0 estimation method)Observed power spectrum in units of centsA band pass filter designed for most melodyObserved pdf of frequency componentsEach observed pdf is generated from weighted-
mixture model of possible tone model Estimate the weighting by EM algorithm (MAP),
regard as F0’s pdfTrack dominant F0
11 cent=
100semi tone
Accompaniment Sound Reduction Harmonic Structure Extraction
Extract the frequency and amplitude of the
l-th overtoneAllow r cent errorSearch local maximum amplitude in an area
Accompaniment Sound Reduction Re-synthesis
Model by sinusoidalQuadratic function approximate changes
in phaseLinear function approximate changes in
amplitude