Polyphonic transcription based on temporal evolution of spectral similarity of gaussian mixture models.
Francisco J. Cañadas-QuesadaPedro Vera-CandeasNicolás Ruiz-ReyesJulio J. Carabias-OrtiPublished in: EUSIPCO (2009)
Keyphrases
- gaussian mixture model
- temporal evolution
- mixture model
- temporal information
- speaker recognition
- similarity measure
- em algorithm
- non stationary
- expectation maximization
- maximum likelihood
- feature vectors
- covariance matrices
- speaker identification
- distance measure
- variational bayes
- feature space
- spectral features
- finite mixtures
- video sequences
- hyperspectral
- density estimation
- speech recognition
- hidden markov models
- probability density
- bayesian information criterion
- covariance matrix