An Investigation of Deep Neural Networks for Multilingual Speech Recognition Training and Adaptation

Different training and adaptation techniques for multilingual Automatic Speech Recognition (ASR) are explored in the context of hybrid systems, exploiting Deep Neural Networks (DNN) and Hidden Markov Models (HMM). In multilingual DNN training, the hidden layers (possibly extracting bottleneck features) are usually shared across languages, and the output layer can either model multiple sets of language-specific senones or one single universal IPA-based multilingual senone set. Both architectures are investigated, exploiting and comparing different language adaptive training (LAT) techniques originating from successful DNN-based speaker-adaptation. More specifically, speaker adaptive training methods such as Cluster Adaptive Training (CAT) and Learning Hidden Unit Contribution (LHUC) are considered. In addition, a language adaptive output architecture for IPA-based universal DNN is also studied and tested. Experiments show that LAT improves the performance and adaptation on the top layer further improves the accuracy. By combining state-level minimum Bayes risk (sMBR) sequence training with LAT, we show that a language adaptively trained IPA-based universal DNN outperforms a monolingually sequence trained model.

Multilingual Training and Adaptation in Speech Recognition

Sibo Tong

State-of-the-art acoustic models for Automatic Speech Recognition (ASR) are based on Hidden Markov Models (HMM) and Deep Neural Networks (DNN) and often require thousands of hours of transcribed speech data during training. Therefore, building multilingual ASR systems or systems on a language with few resources is a challenging task. Multilingual training and cross-lingual adaptation are potential solutions. However, context-dependent states modeling creates difficulties for multilingual and cross-lingual ASR because of the large increase in context dependent labels arising from the phone set mismatch. The goal of this thesis is to improve current state-of-the-art acoustic modeling techniques in general for ASR, with a particular focus on multilingual ASR and cross-lingual adaptation. We systematically exploited new training frameworks, from Maximum Likelihood Estimation, Connectionist Temporal Classification to Maximum Mutual Information, in the context of phoneme-based multilingual training. In order to minimize the negative effects of data impurity arising from language mismatch, we investigated language adaptive training approaches which help further improve the multilingual ASR performance. Through comprehensive experimental comparison we demonstrated that phoneme-based multilingual models are easily extensible to unseen phonemes of new languages, from which the cross-lingual adaptation yields significant improvement over traditional approaches on limited data. Finally, we proposed a semi-supervised training approach based on dropout to boost the performance in low-resourced languages using untranscribed data. In the other part of the thesis, we conducted more theoretical analysis of techniques found to be useful in sequential multilingual training. More specifically, we revisited the recurrent architecture based on Bayesâs theorem. This leads to a Bayesian recurrent unit dictated by the probabilistic formulation and naturally support a backward recursion. Experiments show that the proposed architecture exceeds the performance of conventional recurrent network. Together, this thesis constitutes a thorough analysis of the current field. Through theoretical and experimental comparisons, the proposed approaches are shown to yield significant improvement over the conventional hybrid systems on multilingual speech recognition.

EPFL2020

Neural VTLN for Speaker Adaptation in TTS

Philip Neil Garner, Bastian Schnell

Vocal tract length normalisation (VTLN) is well established as a speaker adaptation technique that can work with very little adaptation data. It is also well known that VTLN can be cast as a linear transform in the cepstral domain. Building on this latter property, we show that it can be cast as a (linear) layer in a deep neural network (DNN) for speech synthesis. We show that VTLN parameters can then be trained in the same framework as the rest of the DNN using automatic gradients. Experimental results show that the DNN is capable of predicting phone-dependent warpings on artificial data, and that such warpings improve the quality of an acoustic model on real data in subjective listening tests.

2019

Exploiting Contextual Information for Improved Phoneme Recognition

Hynek Hermansky, Joel Praveen Pinto

In this paper, we investigate the significance of contextual information in a phoneme recognition system using the hidden Markov model - artificial neural network paradigm. Contextual information is probed at the feature level as well as at the output of the multilayerd perceptron. At the feature level, we analyse and compare different methods to model sub-phonemic classes. To exploit the contextual information at the output of the multilayered perceptron, we propose the hierarchical estimation of phoneme posterior probabilities. The best phoneme (excluding silence) recognition accuracy of 73.4% on the TIMIT database is comparable to that of the state-of-the-art systems, but more emphasis is on analysis of the contextual information.

2008

Multilingual Training and Adaptation in Speech Recognition

Sibo Tong

EPFL2020