Vector quantization of mel-cepstral coefficients using distortion measure for spectral analysis.
Toru TakahashiKeiichi TokudaTakao KobayashiTadashi KitamuraPublished in: Systems and Computers in Japan (2007)
Keyphrases
- cepstral coefficients
- spectral analysis
- distortion measure
- vector quantization
- speech signal
- speech recognition
- linear prediction
- vector quantizer
- image compression
- non stationary
- speaker identification
- hidden markov models
- codebook design
- subband
- filter bank
- autoregressive
- feature set
- computer vision
- speaker recognition
- spectral features
- reference frame
- noisy images
- pattern recognition