A compact model for speaker-adaptive training

Anastasakos, Tasos; McDonough, John; Schwartz, Richard; Makhoul, John

doi:10.1109/icslp.1996.607807

Cited by 393 publications

(296 citation statements)

References 8 publications

Supporting

Mentioning

295

Contrasting

Order By: Relevance

“…In the original formulation (Anastasakos et al, 1996) it involved maximum likelihood linear regression (MLLR) adaptation of the means of output distributions of continuous density HMMs. The resulting HMMs exhibit usually smaller variances and lead to significantly higher likelihood.…”

Section: Satmentioning

confidence: 99%

“…As a more general solution, to tackle inter-speaker spectral variability in children's speech when training on speech from children of all ages, speaker adaptive acoustic modeling methods can be adopted (Giuliani and Gerosa, 2003;Hagen et al, 2004;Giuliani et al, 2006). In this work, we investigated the use of speaker adaptive acoustic modeling methods, such as vocal tract length normalization (VTLN) (Wegmann et al, 1996;Lee and Rose, 1996;Eide and Gish, 1996), constrained MLLR based speaker normalization (CMLSN) (Giuliani et al, 2006), speaker adaptive training (SAT) (Anastasakos et al, 1996;Gales, 1998) and their combinations. These methods proved to be effective in reducing inter-speaker variability and improved recognition performance on children's speech both in matched conditions, that is training and testing on Italian children aged 7-13, and in unmatched conditions, that is testing on children's speech with models trained on adult speech.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Acoustic variability and automatic recognition of children’s speech

Gerosa

Giuliani

Brugnara

2007

Speech Communication

124

View full text Add to dashboard Cite

Section: Satmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Acoustic variability and automatic recognition of children’s speech

Gerosa

Giuliani

Brugnara

2007

Speech Communication

124

View full text Add to dashboard Cite

“…Each constructed transformation W n corresponds to speaker characteristics of a pseudo-speaker n. We reverse the SAT technique [4] by applying the transformation to the normalized speech features to obtain a variety of speakers. We first apply the normalization matrix for training speaker i, W (−1) i , to the speech features of speaker i and then apply the constructed transformation, W n , to them to generate the speech features of pseudo-speaker n:…”

Section: Speech Feature Generation By Mllr Transformations For Pseudomentioning

confidence: 99%

“…Speaker adaptive training (SAT) [4] has also been proManuscript received December 26, 2011. Manuscript revised April 24, 2012.…”

Section: Introductionmentioning

confidence: 99%

Acoustic Model Training Using Pseudo-Speaker Features Generated by MLLR Transformations for Robust Speaker-Independent Speech Recognition

Itoh

Hara

Kitaoka

et al. 2012

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

SUMMARYA novel speech feature generation-based acoustic model training method for robust speaker-independent speech recognition is proposed. For decades, speaker adaptation methods have been widely used. All of these adaptation methods need adaptation data. However, our proposed method aims to create speaker-independent acoustic models that cover not only known but also unknown speakers. We achieve this by adopting inverse maximum likelihood linear regression (MLLR) transformation-based feature generation, and then we train our models using these features. First we obtain MLLR transformation matrices from a limited number of existing speakers. Then we extract the bases of the MLLR transformation matrices using PCA. The distribution of the weight parameters to express the transformation matrices for the existing speakers are estimated. Next, we construct pseudo-speaker transformations by sampling the weight parameters from the distribution, and apply the transformation to the normalized features of the existing speaker to generate the features of the pseudo-speakers. Finally, using these features, we train the acoustic models. Evaluation results show that the acoustic models trained using our proposed method are robust for unknown speakers.

show abstract

“…Current state-of-the-art speech recognition systems are trained on large amounts of speech data, typically collected in a range of acoustic conditions. It is also expected that collecting data in conditions related to the actual target application conditions should improve performance.To address this problem adaptive training has been proposed [5], [6], [7], [8]. These schemes make use of adaptation/compensation transformations during training.…”

mentioning

confidence: 99%

Discriminative adaptive training with VTS and JUD

Flego

Gales

2009

2009 IEEE Workshop on Automatic Speech Recognition &Amp; Understanding

View full text Add to dashboard Cite

Abstract-Adaptive training is a powerful approach for building speech recognition systems on non-homogeneous training data. Recently approaches based on predictive model-based compensation schemes, such as Joint Uncertainty Decoding (JUD) and Vector Taylor Series (VTS), have been proposed. This paper reviews these model-based compensation schemes and relates them to factor-analysis style systems. Forms of Maximum Likelihood (ML) adaptive training with these approaches are described, based on both second-order optimisation schemes and Expectation Maximisation (EM). However, discriminative training is used in many state-of-the-art speech recognition. Hence, this paper proposes discriminative adaptive training with predictive model-compensation approaches for noise robust speech recognition. This training approach is applied to both JUD and VTS compensation with minimum phone error training. A large scale multi-environment training configuration is used and the systems evaluated on a range of in-car collected data tasks.I. INTRODUCTION Speech recognition in noisy environments is a difficult task. There have been a range of solutions proposed to this problem [1], [2], [3]. One approach that has been successfully applied in a range of tasks is model-based compensation. These schemes include Vector Taylor Series (VTS), [1], and Joint Uncertainty Decoding (JUD), [4]. Here a "clean" acoustic model is adapted to be representative of a model trained in the target acoustic condition. To apply these approaches it is necessary to estimate a background noise model and, using a mismatch function that represents the impact of noise on the clean speech, combine this with the clean model parameters. Schemes have previously been proposed that allow the noise model parameters to be robustly trained using Maximum Likelihood (ML) estimation [3], [5] and will not be discussed in this paper. However, it is not always possible to directly train clean model parameters. Current state-of-the-art speech recognition systems are trained on large amounts of speech data, typically collected in a range of acoustic conditions. It is also expected that collecting data in conditions related to the actual target application conditions should improve performance.To address this problem adaptive training has been proposed [5], [6], [7], [8]. These schemes make use of adaptation/compensation transformations during training. The aim is to train a canonical model removing any dependence on speaker or noise condition. Originally proposed to account for the differences between speakers in the training data [6], these schemes have more recently been applied to situations where there are large variations in the background noise [5], [7]. Model-based compensation schemes have been used [5],

show abstract

A compact model for speaker-adaptive training

Cited by 393 publications

References 8 publications

Acoustic variability and automatic recognition of children’s speech

Acoustic variability and automatic recognition of children’s speech

Acoustic Model Training Using Pseudo-Speaker Features Generated by MLLR Transformations for Robust Speaker-Independent Speech Recognition

Discriminative adaptive training with VTS and JUD

Contact Info

Product

Resources

About