Unsupervised Pre-Training of Bidirectional Speech Encoders via Masked Reconstruction

We propose an approach for pre-training speech representations via a masked reconstruction loss. Our pre-trained encoder networks are bidirectional and can therefore be used directly in typical bidirectional speech recognition models. The pre-trained networks can then be fine-tuned on a smaller amount of supervised data for speech recognition. Experiments with this approach on the LibriSpeech and Wall Street Journal corpora show promising results. We find that the main factors that lead to speech recognition improvements are: masking segments of sufficient width in both time and frequency, pre-training on a much larger amount of unlabeled data than the labeled data, and domain adaptation when the unlabeled and labeled data come from different domains. The gain from pre-training is additive to that of supervised data augmentation.

[1]  Yoshua Bengio,et al.  Understanding the difficulty of training deep feedforward neural networks , 2010, AISTATS.

[2]  Jürgen Schmidhuber,et al.  Long Short-Term Memory , 1997, Neural Computation.

[3]  Daniel Povey,et al.  The Kaldi Speech Recognition Toolkit , 2011 .

[4]  Yoshua Bengio,et al.  Learning deep representations by mutual information estimation and maximization , 2018, ICLR.

[5]  Kaisheng Yao,et al.  Adaptation of context-dependent deep neural networks for automatic speech recognition , 2012, 2012 IEEE Spoken Language Technology Workshop (SLT).

[6]  Yu Zhang,et al.  Unsupervised Learning of Disentangled and Interpretable Representations from Sequential Data , 2017, NIPS.

[7]  James R. Glass,et al.  Unsupervised Learning of Spoken Language with Visual Context , 2016, NIPS.

[8]  Tara N. Sainath,et al.  Deep Neural Networks for Acoustic Modeling in Speech Recognition: The Shared Views of Four Research Groups , 2012, IEEE Signal Processing Magazine.

[9]  Hao Tang,et al.  An Unsupervised Autoregressive Model for Speech Representation Learning , 2019, INTERSPEECH.

[10]  Christian Biemann,et al.  Unspeech: Unsupervised Speech Context Embeddings , 2018, INTERSPEECH.

[11]  Jürgen Schmidhuber,et al.  Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks , 2006, ICML.

[12]  Aren Jansen,et al.  A segmental framework for fully-unsupervised large-vocabulary speech recognition , 2016, Comput. Speech Lang..

[13]  Sanjeev Khudanpur,et al.  Librispeech: An ASR corpus based on public domain audio books , 2015, 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[14]  Yoshua Bengio,et al.  Learning Problem-agnostic Speech Representations from Multiple Self-supervised Tasks , 2019, INTERSPEECH.

[15]  Ron J. Weiss,et al.  Unsupervised Speech Representation Learning Using WaveNet Autoencoders , 2019, IEEE/ACM Transactions on Audio, Speech, and Language Processing.

[16]  Sakriani Sakti,et al.  The Zero Resource Speech Challenge 2019: TTS without T , 2019, INTERSPEECH.

[17]  Ming-Wei Chang,et al.  BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding , 2019, NAACL.

[18]  Quoc V. Le,et al.  SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition , 2019, INTERSPEECH.

[19]  Luke S. Zettlemoyer,et al.  Deep Contextualized Word Representations , 2018, NAACL.

[20]  Ronan Collobert,et al.  wav2vec: Unsupervised Pre-training for Speech Recognition , 2019, INTERSPEECH.

[21]  Omer Levy,et al.  GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understanding , 2018, BlackboxNLP@EMNLP.

[22]  Alexei Baevski,et al.  vq-wav2vec: Self-Supervised Learning of Discrete Speech Representations , 2019, ICLR.

[23]  Geoffrey E. Hinton,et al.  Rectified Linear Units Improve Restricted Boltzmann Machines , 2010, ICML.

[24]  Yajie Miao,et al.  EESEN: End-to-end speech recognition using deep RNN models and WFST-based decoding , 2015, 2015 IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU).

[25]  Ilya Sutskever,et al.  Language Models are Unsupervised Multitask Learners , 2019 .

[26]  T. Nilsson,et al.  GPT2: Empirical slant delay model for radio space geodetic techniques , 2013, Geophysical research letters.

[27]  Oriol Vinyals,et al.  Representation Learning with Contrastive Predictive Coding , 2018, ArXiv.

[28]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[29]  Nitish Srivastava,et al.  Dropout: a simple way to prevent neural networks from overfitting , 2014, J. Mach. Learn. Res..

[30]  Geoffrey E. Hinton.  Training Products of Experts by Minimizing Contrastive Divergence , 2002, Neural Computation.