Context-Dependent Phonetic Hidden Markov Models for Speaker-Independent Continuous Speech Recognition

Lee, K.-F.

doi:10.1007/978-3-642-76626-8_15

Cited by 146 publications

(109 citation statements)

References 26 publications

Supporting

Mentioning

108

Contrasting

Unclassified

Order By: Relevance

“…It is empirically found that the proposed LP approach is effective in estimating the stream weights for a multi-stream HMM system either with (1) complete knowledge of the feasible region when it is formulated on frame recognition correctness in a single iteration, or (2) incomplete knowledge of the feasible region when it is formulated on word recognition correctness in several iterations with further constraint on the change in weights. Weights estimated by (1) give the same performance as the "optimal" global weights found by extensive (and computationally expensive) grid search, whereas weights estimated by (2) give even better performance. It is worth noting that in [7], state-dependent stream weights perform worse than global stream weights when they are trained by maximum-entropy estimation.…”

Section: Discussionmentioning

confidence: 87%

“…From another point of view, the slack variables are a measure of frame recognition errors. 1 In this paper, vector quantities are written in bold.…”

Section: The Basic Requirementmentioning

confidence: 99%

“…For instance, in a multi-stream hidden Markov model (HMM) (which is used in discrete HMM system [1,2], multiband ASR [3], and audio-visual ASR [4,5]), the state log-likelihood is usually computed as a linear combination of the per-stream state log-likelihoods; in Viterbi search, the recognition score of a test utterance is a linearly weighted sum of the acoustic score and language score. It is known that the parameter of these linear functions cannot be determined by maximum-likelihood estimation as it will simply give all the weights to the most probable factor.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Discriminative training by iterative linear programming optimization

Mak

2008

2008 IEEE International Conference on Acoustics, Speech and Signal Processing

View full text Add to dashboard Cite

In this paper, we cast discriminative training problems into standard linear programming (LP) optimization. Besides being convex and having globally optimal solution(s), LP programs are well-studied with well-established solutions, and efficient LP solvers are freely available. In practice, however, one may not have complete knowledge of the feasible region since it is constructed from a limited number of competing hypotheses based on the current model -not the final model which, by definition, is not known a priori at the time of hypotheses generation. We investigate an iterative LP optimization algorithm in which an additional constraint on the parameters being optimized is further imposed. Our proposed method is evaluated on the estimation of global and state-dependent stream weights and biases of a multi-stream hidden Markov model system. Results show that the stream weights and biases found by our iterative LP optimization algorithm may give better recognition performance than the ones found by a brute-force grid search.

show abstract

Section: Discussionmentioning

confidence: 87%

“…From another point of view, the slack variables are a measure of frame recognition errors. 1 In this paper, vector quantities are written in bold.…”

Section: The Basic Requirementmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Discriminative training by iterative linear programming optimization

Mak

2008

2008 IEEE International Conference on Acoustics, Speech and Signal Processing

View full text Add to dashboard Cite

show abstract

“…The standard HMMs were estimated as contextdependent models [30] and applied the decision tree based context clustering technique [31]. The minimum description length (MDL) criterion was used to determine the size of the decision trees [32].…”

Section: Methodsmentioning

confidence: 99%

Integration of Spectral Feature Extraction and Modeling for HMM-Based Speech Synthesis

Nakamura

Hashimoto

Nankaku

et al. 2014

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

SUMMARYThis paper proposes a novel approach for integrating spectral feature extraction and acoustic modeling in hidden Markov model (HMM) based speech synthesis. The statistical modeling process of speech waveforms is typically divided into two component modules: the frame-byframe feature extraction module and the acoustic modeling module. In the feature extraction module, the statistical mel-cepstral analysis technique has been used and the objective function is the likelihood of mel-cepstral coefficients for given speech waveforms. In the acoustic modeling module, the objective function is the likelihood of model parameters for given melcepstral coefficients. It is important to improve the performance of each component module for achieving higher quality synthesized speech. However, the final objective of speech synthesis systems is to generate natural speech waveforms from given texts, and the improvement of each component module does not always lead to the improvement of the quality of synthesized speech. Therefore, ideally all objective functions should be optimized based on an integrated criterion which well represents subjective speech quality of human perception. In this paper, we propose an approach to model speech waveforms directly and optimize the final objective function. Experimental results show that the proposed method outperformed the conventional methods in objective and subjective measures. key words: integrative model, HMM-based speech synthesis, acoustic modeling, mel-cepstral analysis, trajectory HMM

show abstract

“…Therefore, the use of the contextual information is very important in modeling characters. Context Dependent Modeling is one of the standard components used in building speech recognition systems [7], [8]. This approach was also successful for handwriting recognition systems.…”

Section: Introductionmentioning

confidence: 99%

Improvement of Context Dependent Modeling for Arabic Handwriting Recognition

Hamdani

Doetsch

Ney

2014

2014 14th International Conference on Frontiers in Handwriting Recognition

View full text Add to dashboard Cite

Abstract-This paper proposes the improvement of context dependent modeling for Arabic handwriting recognition. Since the number of parameters in context dependent models is huge, CART trees are used for state tying. This work is based on a new set of questions for the CART tree construction based on a "lossy mapping" categorization of the Arabic shapes. The used system is a combination of Hidden Markov Models and Recurrent Neural Networks using the hybrid approach. A comparison between a Neural network trained using the baseline labels and another one based on the CART tree labels is done. The experimental results show that the use of the CART labels for the Neural Network training beneficial. The lossy mapping based CART tree performed better than the baseline system. An absolute improvement of 2.9% in terms of Word Error Rate is performed on the test set of the OpenHaRT database.

show abstract

Context-Dependent Phonetic Hidden Markov Models for Speaker-Independent Continuous Speech Recognition

Cited by 146 publications

References 26 publications

Discriminative training by iterative linear programming optimization

Discriminative training by iterative linear programming optimization

Integration of Spectral Feature Extraction and Modeling for HMM-Based Speech Synthesis

Improvement of Context Dependent Modeling for Arabic Handwriting Recognition

Contact Info

Product

Resources

About