Title | ||
---|---|---|
On modeling context-dependent clustered states: Comparing HMM/GMM, hybrid HMM/ANN and KL-HMM approaches |
Abstract | ||
---|---|---|
Deep architectures have recently been explored in hybrid hidden Markov model/artificial neural network (HMM/ANN) framework where the ANN outputs are usually the clustered states of context-dependent phones derived from the best performing HMM/Gaussian mixture model (GMM) system. We can view a hybrid HMM/ANN system as a special case of recently proposed Kullback-Leibler divergence based hidden Markov model (KL-HMM) approach. In KL-HMM approach a probabilistic relationship between the ANN outputs and the context-dependent HMM states is modeled. In this paper, we show that in KL-HMM framework we may not require as many clustered states as the best HMM/GMM system in the ANN output layer. Our experimental results on German part of Media-Parl database show that KL-HMM system achieves better performance compared to hybrid HMM/ANN and HMM/GMM systems with much fewer number of clustered states than is required for HMM/GMM system. The reduction in number of clustered states has broader implications on model complexity and data sparsity issues. |
Year | DOI | Venue |
---|---|---|
2014 | 10.1109/ICASSP.2014.6855090 | Acoustics, Speech and Signal Processing |
Keywords | Field | DocType |
Gaussian processes,hidden Markov models,mixture models,neural nets,speech recognition,HMM-GMM approach,HMM-Gaussian mixture model system,HYBRID HMM-ANN approach,KL-HMM approach,Kullback-Leibler divergence based hidden Markov model approach,Media-Parl database,context-dependent phones,data sparsity,hybrid hidden Markov model-artificial neural network,model complexity,number context-dependent clustered state modeling,HMM/GMM,Kullback-Leibler divergence based HMM,context-dependent subword units,hybrid HMM/ANN,non-native speech recognition | Pattern recognition,Computer science,Speech recognition,Artificial intelligence,Probabilistic logic,Artificial neural network,Hidden Markov model,Mixture model,Model complexity,Special case | Conference |
ISSN | Citations | PageRank |
1520-6149 | 6 | 0.49 |
References | Authors | |
9 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Marzieh Razavi | 1 | 30 | 4.12 |
Ramya Rasipuram | 2 | 57 | 6.90 |
Mathew Magimai-Doss | 3 | 516 | 54.76 |