Vector quantization based on Gaussian mixture models.
Per HedelinJan SkoglundPublished in: IEEE Trans. Speech Audio Process. (2000)
Keyphrases
- vector quantization
- gaussian mixture model
- speaker recognition
- image compression
- mixture model
- vector quantizer
- distortion measure
- feature vectors
- expectation maximization
- maximum likelihood
- em algorithm
- reduced complexity
- probability density
- feature space
- finite state vector quantization
- speaker identification
- density estimation
- wavelet transform
- machine learning
- probability density function
- unsupervised learning
- image processing
- information retrieval
- entropy constrained
- fractal image coding
- codebook design
- successive approximation
- neural network