In this paper, we investigate the effects and problems of MLLR speaker adaptation when applied to pronunciation evaluation. Automatic scoring and error detection experiments are conducted on two publicly available databases of Japanese learners’ English pronunciation. As we expected, overadaptation causes misjudge of pronunciation accuracy. Following these experiments, two novel methods, Forced...

This paper investigates several di erent methods for performing vocal tract length normalisation (VTLN) which are either completely linear or piece-wise linear. Furthermore the combination of VTLN with either standard unconstrained maximum likelihood linear regression (MLLR) or constrained MLLR is considered. Results on the Switchboard corpus show that there is little di erence in performance b...

A variety of methods are used for speaker adaptation in speech recognition. In some techniques, such as MAP estimation, only the models with available training data are updated. Hence, large amounts of training data are required in order to have significant recognition improvements. In some others, such as MLLR, where several general transformations are applied to model clusters, the results ar...

This paper describes a principled application of twodimensional principal component analysis (2DPCA) to the decomposition of transformation matrices of maximum likelihood linear regression (MLLR) and its application to speaker adaptation using the bases derived from the analysis. Our previous work applied 2DPCA to speaker-dependent (SD) models to obtain the bases for state space. In this work, ...

One of the key issues for adaptation algorithms is to modify a large number of parameters with only a small amount of adaptation data. Speaker adaptation techniques try to obtain near speaker dependent (SD) performance with only small amounts of speaker speciic data, and are often based on initial speaker independent (SI) recognition systems. Some of these speaker adaptation techniques may also...

Speaker normalization and speaker adaptation are two strategies to tackle the variations from speaker, channel, and environment. The vocal tract length normalization (VTLN) is an e ective speaker normalization approach to compensate for the variations of vocal tract shapes. The Maximum Likelihood Linear Regression(MLLR) is a recent proposed method for speaker-adaptation. In this paper, we propo...

This paper describes a style adaptation technique using hidden semi-Markov model (HSMM) based maximum likelihood linear regression (MLLR). The HSMM-based MLLR technique can estimate regression matrices for affine transform of mean vectors of output and state duration distributions which maximize likelihood of adaptation data using EM algorithm. In this study, we apply this adaptation technique ...

This paper presents a novel target-driven MLLR adaptation algorithm with multiply layer structure, which is based on the thorough analysis of MLLR using the generation of regression class trees. The new algorithm is constructed on the targetdriven principal. It generates the regression class dynamically, basing on the outcome of the former MLLR transformation. The regression classes is defined ...

In this paper a novel method for speaker adaptive training (SAT), based on Gaussian mean offset adaptation, so called Shift-MLLR, is presented. The method differs from previous SAT methods, where linear transformations of Gaussian means or features are utilized, in that only an offset vector is used for adaptation, but instead the number of regression classes is increased. This is shown to allo...

In this paper we explore several options for cross-task adaptation in speech recognition and compare them to develop the new system from scratch. We compare cross-task MAP and MLLR adaptation, and both of them together, in two speech recognizers for air traffic control tasks, one for spontaneous speech and the other one for a command interface. We show how MLLR can even outperform MAP when a bi...