Title | ||
---|---|---|
Substate tying with combined parameter training and reduction in tied-mixture HMM design |
Abstract | ||
---|---|---|
Two approaches are proposed for the design of tied-mixture hidden Markov models (TMHMM). One approach improves parameter sharing via partial tying of TMHMM states. To facilitate tying at the substate level, the state emission probabilities are constructed in two stages or, equivalently, are viewed as a "mixture of mixtures of Gaussians." This paradigm allows, and is complemented with, an optimization technique to seek the best complexity-accuracy tradeoff solution, which jointly exploits Gaussian density sharing and substate tying. Another approach to enhance model training is combined training and reduction of model parameters. The procedure starts by training a system with a large universal codebook of Gaussian densities. It then iteratively reduces the size of both the codebook and the mixing coefficient matrix, followed by parameter re-training. The additional cost in design complexity is modest. Experimental results on the ISOLET database and its E-set subset show that substate tying reduces the classification error rate by over 15%, compared to standard Gaussian sharing and whole-state tying. TMHMM design with combined training and reduction of parameters reduces the classification error rate by over 20% compared to conventional TMHMM design. When the two proposed approaches were integrated, 25% error rate reduction over TMHMM with whole-state tying was achieved |
Year | DOI | Venue |
---|---|---|
2002 | 10.1109/TSA.2002.1001978 | IEEE Transactions on Speech and Audio Processing |
Keywords | Field | DocType |
tied-mixture hmm design,optimisation,e-set subset,parameter training,state tying,classification error rate reduction,speech recognition,parameter sharing,index terms—parameter reduction,isolet database,codebook size reduction,matrix algebra,state emission probabilities,design complexity,speech recognizers,substate tying,combined parameter training,mixing coefficient matrix,tied-mixture hidden markov models,model parameters reduction,complexity-accuracy tradeoff,optimization,tmhmm,gaussian processes,speech coding,gaussian density sharing,mixture of mixtures of gaussians,tied-mixture hmm.,model training,hidden markov models,parameter re-training,probability,mixture of gaussians,robustness,training data,error rate,automatic speech recognition,hidden markov model,helium,databases,indexing terms | Coefficient matrix,Speech coding,Pattern recognition,Computer science,Word error rate,Tying,Speech recognition,Gaussian,Artificial intelligence,Gaussian process,Hidden Markov model,Codebook | Journal |
Volume | Issue | ISSN |
10 | 3 | 1063-6676 |
Citations | PageRank | References |
6 | 0.65 | 21 |
Authors | ||
2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Liang Gu | 1 | 6 | 0.65 |
Kenneth Rose | 2 | 1261 | 119.38 |