In this paper, a framework for maximum a posteriori (MAP) estimation of hidden Markov models (HMM) is presented. Three key issues of MAP estimation, namely, the choice of prior distribution family, the specification of the parameters of prior densities, and the evaluation of the MAP estimates, are addressed. Using HMM's with Gaussian mixture state observation densities as an example, it is assumed that the prior densities for the HMM parameters can be adequately represented as a product of Dirichlet and normal-Wishart densities. The classical maximum likelihood estimation algorithms, namely, the forward-backward algorithm and the segmental k-means algorithm, are expanded, and MAP estimation formulas are developed. Prior density estimation issues are discussed for two classes of applications/spl minus/parameter smoothing and model adaptation/spl minus/and some experimental results are given illustrating the practical interest of this approach. Because of its adaptive nature, Bayesian learning is shown to serve as a unified approach for a wide range of speech recognition applications. >

Maximum a posteriori estimation for multivariate Gaussian mixture observations of Markov chains

Speech recognition in noisy environments: a survey

This paper describes a new model-based speaker adaptation algorithm called the eigenvoice approach. The approach constrains the adapted model to be a linear combination of a small number of basis vectors obtained offline from a set of reference speakers, and thus greatly reduces the number of free parameters to be estimated from adaptation data. These "eigenvoice" basis vectors are orthogonal to each other and guaranteed to represent the most important components of variation between the reference speakers. Experimental results for a small-vocabulary task (letter recognition) given in the paper show that the approach yields major improvements in performance for tiny amounts of adaptation data. For instance, we obtained 16% relative improvement in error rate with one letter of supervised adaptation data, and 26% relative improvement with four letters of supervised adaptation data. After a comparison of the eigenvoice approach with other speaker adaptation algorithms, the paper concludes with a discussion of future work.

Rapid speaker adaptation in eigenvoice space

For a speech-recognition system based on continuous-density hidden Markov models (CDHMM), speaker adaptation of the parameters of CDHMM is formulated as a Bayesian learning procedure. A speaker adaptation procedure which is easily integrated into the segmental k-means training procedure for obtaining adaptive estimates of the CDHMM parameters is presented. Some results for adapting both the mean and the diagonal covariance matrix of the Gaussian state observation densities of a CDHMM are reported. The results from tests on a 39-word English alpha-digit vocabulary in isolated word mode indicate that the speaker adaptation procedure achieves the same level of performance as that of a speaker-independent system, when one training token from each word is used to perform speaker adaptation. It shows that much better performance is achieved when two or more training tokens are used for speaker adaptation. When compared with the speaker-dependent system, it is found that the performance of speaker adaptation is always equal to or better than that of speaker-dependent training using the same amount of training data. >

A study on speaker adaptation of the parameters of continuous density hidden Markov models

: This thesis examines the acoustic-modeling problem in automatic speech recognition from an information-theoretic point of view. This problem is to design a speech-recognition system which can extract from the speech waveform as much information as possible about the corresponding word sequence. The information extraction process is broken down into two steps: a signal processing step which converts a speech waveform into a sequence of information bearing acoustic feature vectors, and a step which models such a sequence. This thesis is primarily concerned with the use of hidden Markov models to model sequences of feature vectors which lie in a continuous space such as R sub N. It explores the trade-off between packing a lot of information into such sequences and being able to model them accurately. The difficulty of developing accurate models of continuous parameter sequences is addressed by investigating a method of parameter estimation which is specifically designed to cope with inaccurate modeling assumptions.

https://apps.dtic.mil/sti/pdfs/ADA188529.pdf

The acoustic-modeling problem in automatic speech recognition

In this paper, we describe efforts to improve the performance of FEATURE, the Carnegie-Mellon University speaker-independent speech recognition system that classifies isolated letters of the English alphabet by enabling the system to learn the acoustical characteristics of individual speakers. Even when features are designed to be speaker-independent, it is frequently observed that feature values may vary more from speaker to speaker for a single letter than they vary from letter to letter. In these cases, it is necessary to adjust the system's statistical description of the features of individual speakers to obtain improved recognition performance. This paper describes a set of dynamic adaptation procedures for updating expected feature values during recognition. The algorithm uses maximum a posteriori probability (MAP) estimation techniques to update the mean vectors of sets of feature values on a speaker-by-speaker basis. The MAP estimation algorithm makes use of both knowledge of the observations input to the system from an individual speaker and the relative variability of the features' means within and across all speakers. In addition, knowledge of the covariance of the features' mean vectors across the various letters enables the system to adapt its representation of similar-sounding letters after any one of them is presented to the classifier. The use of dynamic speaker adaptation improves classification performance of FEATURE by 49 percent after four presentations of the alphabet, when the system is provided with supervised training indicating which specific utterance had been presented to the classifier from a particular user. Performance can be improved by as much as 31 percent when the system is allowed to adapt passively in an unsupervised learning mode. without any information from individual users.

Dynamic speaker adaptation for feature-based isolated word recognition

This paper describes the use of maximum a posteriori probability (MAP) techniques to estimate the mean values of features used in statistical pattern classification problems, when these mean feature values from the various decision classes are jointly Gaussian random vectors that are correlated across the decision classes. A set of mathematical formalisms is proposed and used to derive closed-form expressions for the estimates of the class-conditional mean vectors, and for the covariance matrix of the errors of these estimates. Finally, the performance of these algorithms is described for the simple case of a two-class one-feature pattern recognition problem, and compared to the performance of classical estimators that do not exploit the class-to-class correlations of the features' mean values.

/pdf/a-posteriori-estimation-of-correlated-jointly-gaussian-mean-g3vpizp9qb.pdf

A Posteriori Estimation of Correlated Jointly Gaussian Mean Vectors

Performing fine phonetic distinctions: templates versus features

A dynamic speaker-adaptation algorithm for the C-MU feature-based isolated letter recognition system, FEATURE, is described. The algorithm, based on maximum a posteriori probability estimation techniques, uses the labelled observations input thus far to the classifier, as well as the a priori correlations of the features within and across the various letters or sets of letters (classes). The probability density functions (pdf) of all the classes are updated simultaneously rather than on a class-by-class basis so that the pdf of a given class is updated before any observation from that class has been input. A significant improvement in the recognition performance was observed for different vocabularies as the system tuned to the the characteristics of a new speaker. Finally, the algorithm was compared to simpler forms of dynamic adaptation. It produced a faster decrease of the error rate than the other tuning procedures. After a small number of iterations, however, the various procedures yielded similar results.

Dynamic speaker adaptation for isolated letter recognition using MAP estimation

This paper describes two new methods by which the CMU feature-based recognition system can learn the acoustical characteristics of individual speakers without feedback from the user. We have previously described how the system uses MAP techniques to update its estimates of the mean values of features used by the classifier in recognizing the letters of the English alphabet on the basis of a priori information and labelled observations. In the first of the new procedures described in this paper the system assumes a correct decision every time it classifies a new utterance with a sufficiently high confidence level. In the second new procedure the system adjusts its estimates of the means on the basis of their correlation with the average values of the features over all utterances. Experiments were conducted on two confusable sets of letters using both speaker adaptation procedures. In each case classification performance using the unsupervised estimation procedures could equal that obtained using speaker adaptation with feedback from the user, although which method provided the better performance depended on which set of letters was being classified.

Moshe J. Lasry

Papers

Dynamic speaker adaptation for feature-based isolated word recognition

A Posteriori Estimation of Correlated Jointly Gaussian Mean Vectors

Performing fine phonetic distinctions: templates versus features

Dynamic speaker adaptation for isolated letter recognition using MAP estimation

Unsupervised adaptation to new speakers in feature-based letter recognition