Employment of Subspace Gaussian Mixture Models in speaker recognition.
Petr MotlícekSubhadeep DeySrikanth R. MadikeriLukás BurgetPublished in: ICASSP (2015)
Keyphrases
- speaker recognition
- gaussian mixture model
- feature space
- mixture model
- speaker identification
- principal component analysis
- high dimensional
- low dimensional
- feature vectors
- mel frequency cepstral coefficients
- dimensionality reduction
- maximum likelihood
- vector quantization
- high dimensional data
- em algorithm
- expectation maximization
- speaker verification
- neural network
- image compression
- active learning
- feature extraction