Class-Discriminative Weighted Distortion Measure for VQ-based Speaker Identification.
Tomi KinnunenIsmo KärkkäinenPublished in: SSPR/SPR (2002)
Keyphrases
- distortion measure
- vector quantization
- speaker identification
- vector quantizer
- image coding
- image compression
- human perception
- rate distortion
- gaussian mixture model
- feature extraction
- speech recognition
- noisy environments
- bit allocation
- spatial domain
- high level
- structural similarity
- speech signal
- feature vectors
- human visual system
- image processing
- signal processing
- unsupervised learning