wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations

Alexei Baevski

[0]

Henry Zhou

[0]

Abdelrahman Mohamed

[0]

Michael Auli

[0]

NeurIPS, 2020.

Cited by: 19|Bibtex|Views572

Other Links: arxiv.org|dblp.uni-trier.de|academic.microsoft.com

Keywords:

unsupervised pre trainingConnectionist Temporal Classificationspeech representationlanguage modellatent representationMore(7+)

Weibo:

We presented wav2vec 2.0, a framework for self-supervised learning of speech representations which masks latent representations of the raw waveform and solves a contrastive task over quantized speech representations

Abstract:

We show for the first time that learning powerful representations from speech audio alone followed by fine-tuning on transcribed speech can outperform the best semi-supervised methods while being conceptually simpler. wav2vec 2.0 masks the speech input in the latent space and solves a contrastive task defined over a quantization of the ...More

Code:

https://github.com/pytorch/fairseq

Data:

Insight

Full Text

PPT (Upload PPT)

Introduction

Neural networks benefit from large quantities of labeled training data. in many settings labeled data is much harder to come by than unlabeled data: current speech recognition systems require thousands of hours of transcribed speech to reach acceptable performance which is not available for the vast majority of the nearly 7,000 languages spoken worldwide [30].
Self-supervised learning has emerged as a paradigm to learn general data representations from unlabeled examples and to fine-tune the model on labeled data
This has been successful for natural language processing [42, 44, 9] and is an active research area for computer vision [19, 2, 35, 18, 6].
The latent representations are fed to a Transformer network to build contextualized representations and the model is trained via a contrastive task where the true latent is to be distinguished from distractors [51, 47, 46, 27] (§ 2)

Highlights

Neural networks benefit from large quantities of labeled training data
The latent representations are fed to a Transformer network to build contextualized representations and the model is trained via a contrastive task where the true latent is to be distinguished from distractors [51, 47, 46, 27] (§ 2)
Our results demonstrate the feasibility of ultra-low resource speech recognition: when using only 10 minutes of labeled data, our approach achieves word error rate (WER) 5.7/10.1 on the clean/noisy test sets of Librispeech
The models are pre-trained on the audio data of either Librispeech (LS-960) or LibriVox (LV-60k) and most results are obtained by decoding with a Transformer language model (Transf.); Appendix C shows results with other language models
We presented wav2vec 2.0, a framework for self-supervised learning of speech representations which masks latent representations of the raw waveform and solves a contrastive task over quantized speech representations
Our experiments show the large potential of pre-training on unlabeled data for speech processing: when using only 10 minutes of labeled training data, or 48 recordings of 12.5 seconds on average, we achieve a WER of 5.7/10.1 on test-clean/other of Librispeech

Methods

As unlabeled data the authors consider the Librispeech corpus [39] without transcriptions containing 960 hours of audio (LS-960) or the audio data from LibriVox (LV-60k).
For the latter the authors follow the preprocessing of [26] resulting in 53.2k hours of audio.
The authors fine-tune the pre-trained models for phoneme recognition on the TIMIT dataset [13]
It contains five hours of audio recordings with detailed phoneme labels.
The authors use the standard train, dev and test split and follow the standard protocol of collapsing phone labels to 39 classes

Results

The authors first evaluate the pre-trained models in settings where the amount of labeled data is limited to get a sense of how the representations learned on unlabeled data can improve low resource settings.
The LARGE model pre-trained on LV-60k and fine-tuned on only 10 minutes of labeled data achieves a word error rate of 5.7/10.1 on the Librispeech clean/other test sets.
Ten minutes of labeled data corresponds to just 48 recordings with an average length of 12.5 seconds
This demonstrates that ultra-low resource speech recognition is possible with self-supervised learning on unlabeled data.
The authors' approach improves over previous pre-training work which did not learn quantized audio units jointly [4], reducing WER by a about a third

Conclusion

The authors presented wav2vec 2.0, a framework for self-supervised learning of speech representations which masks latent representations of the raw waveform and solves a contrastive task over quantized speech representations.
The authors' experiments show the large potential of pre-training on unlabeled data for speech processing: when using only 10 minutes of labeled training data, or 48 recordings of 12.5 seconds on average, the authors achieve a WER of 5.7/10.1 on test-clean/other of Librispeech.
The authors' model achieves a new state of the art on the clean 100 hour Librispeech setup and outperforms the previous best result even when using 100 times less labeled data.
The approach is effective when large amounts of labeled data are available.

Summary

Introduction:
Neural networks benefit from large quantities of labeled training data. in many settings labeled data is much harder to come by than unlabeled data: current speech recognition systems require thousands of hours of transcribed speech to reach acceptable performance which is not available for the vast majority of the nearly 7,000 languages spoken worldwide [30].
Self-supervised learning has emerged as a paradigm to learn general data representations from unlabeled examples and to fine-tune the model on labeled data
This has been successful for natural language processing [42, 44, 9] and is an active research area for computer vision [19, 2, 35, 18, 6].
The latent representations are fed to a Transformer network to build contextualized representations and the model is trained via a contrastive task where the true latent is to be distinguished from distractors [51, 47, 46, 27] (§ 2)
Methods:
As unlabeled data the authors consider the Librispeech corpus [39] without transcriptions containing 960 hours of audio (LS-960) or the audio data from LibriVox (LV-60k).
For the latter the authors follow the preprocessing of [26] resulting in 53.2k hours of audio.
The authors fine-tune the pre-trained models for phoneme recognition on the TIMIT dataset [13]
It contains five hours of audio recordings with detailed phoneme labels.
The authors use the standard train, dev and test split and follow the standard protocol of collapsing phone labels to 39 classes
Results:
The authors first evaluate the pre-trained models in settings where the amount of labeled data is limited to get a sense of how the representations learned on unlabeled data can improve low resource settings.
The LARGE model pre-trained on LV-60k and fine-tuned on only 10 minutes of labeled data achieves a word error rate of 5.7/10.1 on the Librispeech clean/other test sets.
Ten minutes of labeled data corresponds to just 48 recordings with an average length of 12.5 seconds
This demonstrates that ultra-low resource speech recognition is possible with self-supervised learning on unlabeled data.
The authors' approach improves over previous pre-training work which did not learn quantized audio units jointly [4], reducing WER by a about a third
Conclusion:
The authors presented wav2vec 2.0, a framework for self-supervised learning of speech representations which masks latent representations of the raw waveform and solves a contrastive task over quantized speech representations.
The authors' experiments show the large potential of pre-training on unlabeled data for speech processing: when using only 10 minutes of labeled training data, or 48 recordings of 12.5 seconds on average, the authors achieve a WER of 5.7/10.1 on test-clean/other of Librispeech.
The authors' model achieves a new state of the art on the clean 100 hour Librispeech setup and outperforms the previous best result even when using 100 times less labeled data.
The approach is effective when large amounts of labeled data are available.

Tables

Table1: WER on the Librispeech dev/test sets when training on the Libri-light low-resource labeled data setups of 10 min, 1 hour, 10 hours and the clean 100h subset of Librispeech. Models use either the audio of Librispeech (LS-960) or the larger LibriVox (LV-60k) as unlabeled data. We consider two model sizes: BASE (95m parameters) and LARGE (317m parameters). Prior work used 860 unlabeled hours (LS-860) but the total with labeled data is 960 hours and comparable to our setup
Table2: WER on Librispeech when using all labeled data of 960 hours (cf
Table3: TIMIT phoneme recognition accuracy in terms of phoneme error rate (PER)
Table4: Average WER and standard deviation on combined dev-clean/other of Librispeech for three training seeds. We ablate quantizing the context network input and the targets in the contrastive loss
Table5: Ablations on settings for the masking strategy during pre-training. When masking without overlap, we choose starting time steps with p = 0.037 which results in the total number of masked tokens to match the baseline
Table6: Fine-tuning hyperparameters timestep mask prob. channel mask prob
Table7: Decoding parameters for Librispeech subsets
Table8: WER on the Librispeech dev/test sets when training on the Libri-light low-resource labeled data setups (cf
Table9: WER on Librispeech when using all 960 hours of Librispeech as labeled data (cf
Table10: Top word errors for models trained on 10m, 1h and 10h, 100h, 960h of labeled data and decoded on the Librispeech dev-clean subset without a language model or lexicon (see Table 8 and Table 9 - None). In brackets is the total number of occurrences of each error
Table11: Examples of transcription of selected utterances from the dev-clean subset by various models without a language model or lexicon. Capitalized words indicate errors
Table12: Ablation of various hyper-parmeter choices. We report average WER and standard deviation on combined dev-clean/other of Librispeech for three seeds of training

Download tables as Excel

Reference

J. L. Ba, J. R. Kiros, and G. E. Hinton. Layer normalization. arXiv, 2016.
P. Bachman, R. D. Hjelm, and W. Buchwalter. Learning representations by maximizing mutual information across views. In Proc. of NeurIPS, 2019.
A. Baevski and M. Auli. Adaptive input representations for neural language modeling. In Proc.
A. Baevski, M. Auli, and A. Mohamed. Effectiveness of self-supervised pre-training for speech recognition. arXiv, abs/1911.03912, 2019.
A. Baevski, S. Schneider, and M. Auli. vq-wav2vec: Self-supervised learning of discrete speech representations. In Proc. of ICLR, 2020.
T. Chen, S. Kornblith, M. Norouzi, and G. Hinton. A simple framework for contrastive learning of visual representations. arXiv, abs/2002.05709, 2020.
J. Chorowski, R. J. Weiss, S. Bengio, and A. van den Oord. Unsupervised speech representation learning using wavenet autoencoders. arXiv, abs/1901.08810, 2019.
Y. Chung, W. Hsu, H. Tang, and J. R. Glass. An unsupervised autoregressive model for speech representation learning. arXiv, abs/1904.03240, 2019.
J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova. Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv, abs/1810.04805, 2018.
S. Dieleman, A. van den Oord, and K. Simonyan. The challenge of realistic music generation: modelling raw audio at scale. arXiv, 2018.
R. Eloff, A. Nortje, B. van Niekerk, A. Govender, L. Nortje, A. Pretorius, E. Van Biljon, E. van der Westhuizen, L. van Staden, and H. Kamper. Unsupervised acoustic unit discovery for speech synthesis using discrete latent-variable neural networks. arXiv, abs/1904.07556, 2019.
A. Fan, E. Grave, and A. Joulin. Reducing transformer depth on demand with structured dropout. In Proc. of ICLR, 2020.
J. S. Garofolo, L. F. Lamel, W. M. Fisher, J. G. Fiscus, D. S. Pallett, and N. L. Dahlgren. The DARPA TIMIT Acoustic-Phonetic Continuous Speech Corpus CDROM. Linguistic Data Consortium, 1993.
A. Graves, S. Fernández, and F. Gomez. Connectionist temporal classification: Labelling unsegmented sequence data with recurrent neural networks. In Proc. of ICML, 2006.
E. J. Gumbel. Statistical theory of extreme values and some practical applications: a series of lectures, volume 33. US Government Printing Office, 1954.
W. Han, Z. Zhang, Y. Zhang, J. Yu, C.-C. Chiu, J. Qin, A. Gulati, R. Pang, and Y. Wu. Contextnet: Improving convolutional neural networks for automatic speech recognition with global context. arXiv, 2020.
D. Harwath, W.-N. Hsu, and J. Glass. Learning hierarchical discrete linguistic units from visually-grounded speech. In Proc. of ICLR, 2020.
K. He, H. Fan, Y. Wu, S. Xie, and R. Girshick. Momentum contrast for unsupervised visual representation learning. arXiv, abs/1911.05722, 2019.
O. J. Hénaff, A. Razavi, C. Doersch, S. M. A. Eslami, and A. van den Oord. Data-efficient image recognition with contrastive predictive coding. arXiv, abs/1905.09272, 2019.
D. Hendrycks and K. Gimpel. Gaussian error linear units (gelus). arXiv, 2016.
G. Huang, Y. Sun, Z. Liu, D. Sedra, and K. Weinberger. Deep networks with stochastic depth. arXiv, 2016.
M. G. A. Hyvärinen. Noise-contrastive estimation: A new estimation principle for unnormalized statistical models. In Proc. of AISTATS, 2010.
E. Jang, S. Gu, and B. Poole. Categorical reparameterization with gumbel-softmax. arXiv, abs/1611.01144, 2016.
H. Jegou, M. Douze, and C. Schmid. Product quantization for nearest neighbor search. IEEE Trans. Pattern Anal. Mach. Intell., 33(1):117–128, Jan. 2011.
D. Jiang, X. Lei, W. Li, N. Luo, Y. Hu, W. Zou, and X. Li. Improving transformer-based speech recognition using unsupervised pre-training. arXiv, abs/1910.09932, 2019.
J. Kahn et al. Libri-light: A benchmark for asr with limited or no supervision. In Proc. of ICASSP, 2020.
K. Kawakami, L. Wang, C. Dyer, P. Blunsom, and A. van den Oord. Learning robust and multilingual speech representations. arXiv, 2020.
D. P. Kingma and J. Ba. Adam: A Method for Stochastic Optimization. In Proc. of ICLR, 2015.
A. Laptev, R. Korostik, A. Svischev, A. Andrusenko, I. Medennikov, and S. Rybin. You do not need more data: Improving end-to-end speech recognition by text-to-speech data augmentation. arXiv, abs/2005.07157, 2020.
M. P. Lewis, G. F. Simon, and C. D. Fennig. Ethnologue: Languages of the world, nineteenth edition. Online version: http://www.ethnologue.com, 2016.
A. H. Liu, T. Tu, H. yi Lee, and L. shan Lee. Towards unsupervised speech recognition and synthesis with quantized speech representation learning. arXiv, 2019.
Y. Liu, M. Ott, N. Goyal, J. Du, M. Joshi, D. Chen, O. Levy, M. Lewis, L. Zettlemoyer, and V. Stoyanov. Roberta: A robustly optimized bert pretraining approach. arXiv preprint arXiv:1907.11692, 2019.
C. Lüscher, E. Beck, K. Irie, M. Kitza, W. Michel, A. Zeyer, R. Schlüter, and H. Ney. Rwth asr systems for librispeech: Hybrid vs attention. In Interspeech 2019, 2019.
C. J. Maddison, D. Tarlow, and T. Minka. A* sampling. In Advances in Neural Information Processing Systems, pages 3086–3094, 2014.
I. Misra and L. van der Maaten. Self-supervised learning of pretext-invariant representations. arXiv, 2019.
A. Mohamed, D. Okhonko, and L. Zettlemoyer. Transformers with convolutional context for ASR. arXiv, abs/1904.11660, 2019.
M. Ott, S. Edunov, D. Grangier, and M. Auli. Scaling neural machine translation. In Proc. of WMT, 2018.
M. Ott, S. Edunov, A. Baevski, A. Fan, S. Gross, N. Ng, D. Grangier, and M. Auli. fairseq: A fast, extensible toolkit for sequence modeling. In Proc. of NAACL System Demonstrations, 2019.
V. Panayotov, G. Chen, D. Povey, and S. Khudanpur. Librispeech: an asr corpus based on public domain audio books. In Proc. of ICASSP, pages 5206–5210. IEEE, 2015.
D. S. Park, W. Chan, Y. Zhang, C.-C. Chiu, B. Zoph, E. D. Cubuk, and Q. V. Le. Specaugment: A simple data augmentation method for automatic speech recognition. In Proc. of Interspeech, 2019.
D. S. Park, Y. Zhang, Y. Jia, W. Han, C.-C. Chiu, B. Li, Y. Wu, and Q. V. Le. Improved noisy student training for automatic speech recognition. arXiv, abs/2005.09629, 2020.
M. E. Peters, M. Neumann, M. Iyyer, M. Gardner, C. Clark, K. Lee, and L. Zettlemoyer. Deep contextualized word representations. In Proc. of ACL, 2018.
V. Pratap, A. Hannun, Q. Xu, J. Cai, J. Kahn, G. Synnaeve, V. Liptchinsky, and R. Collobert. Wav2letter++: A fast open-source speech recognition system. In Proc. of ICASSP, 2019.
A. Radford, K. Narasimhan, T. Salimans, and I. Sutskever. Improving language understanding by generative pre-training. https://s3-us-west-2.amazonaws.com/openai-assets/research-covers/language-unsupervised/language_understanding_paper.pdf, 2018.
M. Ravanelli, P. Brakel, M. Omologo, and Y. Bengio. Light gated recurrent units for speech recognition. IEEE Transactions on Emerging Topics in Computational Intelligence, 2(2):92–102, 2018.
M. Rivière, A. Joulin, P.-E. Mazaré, and E. Dupoux. Unsupervised pretraining transfers well across languages. arXiv, abs/2002.02848, 2020.
S. Schneider, A. Baevski, R. Collobert, and M. Auli. wav2vec: Unsupervised pre-training for speech recognition. In Proc. of Interspeech, 2019.
M. Schuster and K. Nakajima. Japanese and korean voice search. In Proc. of ICASSP, 2012.
G. Synnaeve, Q. Xu, J. Kahn, T. Likhomanenko, E. Grave, V. Pratap, A. Sriram, V. Liptchinsky, and R. Collobert. End-to-end ASR: from Supervised to Semi-Supervised Learning with Modern Architectures. arXiv, abs/1911.08460, 2020.
A. Tjandra, B. Sisman, M. Zhang, S. Sakti, H. Li, and S. Nakamura. Vqvae unsupervised unit discovery and multi-scale code2spec inverter for zerospeech challenge 2019. arXiv, 1905.11449, 2019.
A. van den Oord, O. Vinyals, et al. Neural discrete representation learning. In Advances in Neural Information Processing Systems, pages 6306–6315, 2017.
A. van den Oord, Y. Li, and O. Vinyals. Representation learning with contrastive predictive coding. arXiv, abs/1807.03748, 2018.
A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser, and I. Polosukhin. Attention is all you need. In Proc. of NIPS, 2017.
W. Wang, Q. Tang, and K. Livescu. Unsupervised pre-training of bidirectional speech encoders via masked reconstruction. arXiv, 2020.
F. Wu, A. Fan, A. Baevski, Y. N. Dauphin, and M. Auli. Pay less attention with lightweight and dynamic convolutions. In Proc. of ICLR, 2019.
Q. Xu, T. Likhomanenko, J. Kahn, A. Hannun, G. Synnaeve, and R. Collobert. Iterative pseudo-labeling for speech recognition. arXiv, 2020.
N. Zeghidour, N. Usunier, I. Kokkinos, T. Schaiz, G. Synnaeve, and E. Dupoux. Learning filterbanks from raw speech for phone recognition. In Proc. of ICASSP, 2018.
Q. Zhang, H. Lu, H. Sak, A. Tripathi, E. McDermott, S. Koo, and S. Kumar. Transformer transducer: A streamable speech recognition model with transformer encoders and rnn-t loss. arXiv, 2020.

Full Text

Your rating :

wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations

Introduction:

Methods:

Results:

Conclusion: