Fast speaker adaptation combined with soft vector quantization in an HMM speech recognition system.
Fritz ClassAlfred KaltenmeierPeter Regel-BrietzmannKarl TrottlerPublished in: ICASSP (1992)
Keyphrases
- vector quantization
- speaker adaptation
- speech recognition
- image compression
- hidden markov models
- maximum likelihood
- automatic speech recognition
- vector quantizer
- speaker independent
- distortion measure
- codebook design
- reduced complexity
- fractal image coding
- speaker recognition
- finite state vector quantization
- speech recognizer
- image processing
- noisy environments
- rate distortion
- vocal tract
- image quality
- multiresolution
- information retrieval
- tree structured vector quantization
- machine learning