Paper Info
Title
Speaker Adaptation Using Nonlinear Regression Techniques for HMM-Based Speech Synthesis
Abstract
The maximum likelihood linear regression (MLLR) technique is a well-known approach to parameter adaptation in hidden Markov model (HMM)-based systems. In this paper, we propose the maximum penalized likelihood kernel regression (MPLKR) approach as a novel adaptation technique for HMM-based speech synthesis. The proposed algorithm performs a nonlinear regression between the mean vector of the base model and the corresponding mean vector of adaptive data by means of a kernel method. In the experiments, we used various types of parametric kernels for the proposed algorithm and compared their performances with the conventional method. From experimental results, it has been found that the proposed algorithm outperforms the conventional method in terms of the objective measure as well as the subjective listening quality.
Year
DOI
Venue
2014
10.1109/IIH-MSP.2014.152
IIH-MSP
Keywords
Field
DocType
kernel,maximum likelihood linear regression (mllr),subjective listening quality,regression analysis,maximum likelihood estimation,hmm,maximum likelihood linear regression , hmm-based speech synthesis, kernel, maximum penalized likelihood kernel regression,mplkr approach,speech synthesis,parametric kernels,hmm-based speech synthesis,hidden markov model,nonlinear regression techniques,hidden markov models,speaker adaptation,maximum penalized likelihood kernel regression (mplkr),maximum penalized likelihood kernel regression,mean vector
Kernel (linear algebra),Pattern recognition,Principal component regression,Computer science,Nonlinear regression,Speech recognition,Artificial intelligence,Maximum likelihood sequence estimation,Kernel method,Hidden Markov model,Variable kernel density estimation,Kernel regression
Conference
Citations 
PageRank 
References 
0
0.34
8
Authors
4
Name
Order
Citations
PageRank
Doo Hwa Hong1164.55
Shin Jae Kang2556.48
Joun Yeop Lee302.37
Nam Soo Kim416924.18