On the reduction of false positives in singing voice detection
Authors:
- 1 May 2014
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 15206149,p. 7480-7484
Abstract
Motivated by the observation that one of the biggest problems in automatic singing voice detection is the confusion of vocals with other pitch-continuous and pitch-varying instruments, we propose a set of three new audio features designed to reduce the amount of false vocal detections. This is borne out in comparative experiments with three different musical corpora. The resulting singing voice detector appears to be at least on par with more complex state-of-the-art methods. New features and classifier are very light-weight and in principle suitable for on-line use.Keywords
This publication has 6 references indexed in Scilit:
- The WEKA data mining softwareACM SIGKDD Explorations Newsletter, 2009
- Cutting-plane training of structural SVMsMachine Learning, 2009
- Vocal detection in music with support vector machines2008
- YIN, a fundamental frequency estimator for speech and musicThe Journal of the Acoustical Society of America, 2002
- Random ForestsMachine Learning, 2001
- A spectral-flatness measure for studying the autocorrelation method of linear prediction of speech analysisIEEE Transactions on Acoustics, Speech, and Signal Processing, 1974