Comparing maximum a posteriori vector quantization and Gaussian mixture models in speaker verification.
Tomi KinnunenJuhani SaastamoinenVille HautamäkiMikko VinniPasi FräntiPublished in: ICASSP (2009)
Keyphrases
- speaker recognition
- maximum a posteriori
- vector quantization
- gaussian mixture model
- speaker verification
- maximum likelihood
- em algorithm
- expectation maximization
- markov random field
- image compression
- image reconstruction
- vector quantizer
- mixture model
- generalized gaussian
- noisy environments
- audio visual
- speaker identification
- high quality
- multiresolution