On Generative Spoken Language Modeling from Raw Audio
Open accessThe DOI matches its OpenAlex record; 16 of the 16 references it cites were found in a registry or an index.
Abstract
We introduce Generative Spoken Language Modeling, the task of learning the acoustic and linguistic characteristics of a language from raw audio (no text, no labels), and a set of metrics to automatically evaluate the learned representations at acoustic and linguistic levels for both encoding and generation. We set up baseline systems consisting of a discrete speech encoder (returning pseudo-text units), a generative language model (trained on pseudo- text), and a speech decoder (generating a waveform from pseudo-text) all trained without supervision and validate the proposed metrics with human evaluation. Across 3 speech encoders (CPC, wav2vec 2.0, HuBERT), we find that the number of discrete units (50, 100, or 200) matters in a task-dependent and encoder- dependent way, and that some combinations approach text-based systems.1
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325
RoBERTa: A Robustly Optimized BERT Pretraining Approach
Liu, Y., Ott, M., Goyal, N., Du, J., Joshi, M., Chen, D., Levy, O., Lewis, M., Zettlemoyer, L., & Stoyanov, V. (2019) · arXiv · doi:10.48550/arxiv.1907.11692 · Cited by 16,515
Deep Contextualized Word Representations
Peters, M., Neumann, M., Iyyer, M., Gardner, M., Clark, C., Lee, K., & Zettlemoyer, L. (2018) · Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers) · doi:10.18653/v1/n18-1202 · Cited by 11,983
Lewis, M., Liu, Y., Goyal, N., Ghazvininejad, M., Mohamed, A., Levy, O., Stoyanov, V., & Zettlemoyer, L. (2020) · Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/2020.acl-main.703 · Cited by 8,821
Self-citationLewis, M., Liu, Y., Goyal, N., Ghazvininejad, M., Mohamed, A., Levy, O., Stoyanov, V., & Zettlemoyer, L. (2019) · arXiv · doi:10.48550/arxiv.1910.13461 · Cited by 8,816 · cited 1×
“…training objectives (Brown et al., 2020; Peters et al., 2018; Rad-ford et al., 2019; Devlin et al., 2019; Liu et al., 2019b; Dong et al., 2019; Lewis et al., 2020), the basic units on which these systems are trained are https://github.com/pytorch/fairseq/tree/…”
Self-citationRepresentation Learning with Contrastive Predictive Coding
Oord, A. V. D., Li, Y., & Vinyals, O. (2018) · arXiv · doi:10.48550/arxiv.1807.03748 · Cited by 6,242
“Librispeech: an ASR corpus based on public domain audio books,”
Panayotov, V., Chen, G., Povey, D., & Khudanpur, S. (2015) · 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) · doi:10.1109/icassp.2015.7178964 · Cited by 6,072
Graves, A., Fernández, S., Gomez, F., & Schmidhuber, J. (2006) · Proceedings of the 23rd international conference on Machine learning - ICML '06 · doi:10.1145/1143844.1143891 · Cited by 5,370
WaveNet: a generative model for raw audio
Oord, A. V. D., Dieleman, S., Zen, H., Simonyan, K., Vinyals, O., Graves, A., Kalchbrenner, N., Senior, A., & Kavukcuoglu, K. (2016) · arXiv · doi:10.48550/arxiv.1609.03499 · Cited by 4,911
Neural Discrete Representation Learning
Oord, A. V. D., Vinyals, O., & Kavukcuoglu, K. (2017) · arXiv · doi:10.48550/arxiv.1711.00937 · Cited by 2,729
Deep Clustering for Unsupervised Learning of Visual Features
Caron, M., Bojanowski, P., Joulin, A., & Douze, M. (2018) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-01264-9_9 · Cited by 2,674
wav2vec: Unsupervised Pre-Training for Speech Recognition
Schneider, S., Baevski, A., Collobert, R., & Auli, M. (2019) · Interspeech 2019 · doi:10.21437/interspeech.2019-1873 · Cited by 1,290
Self-citationUnified Language Model Pre-training for Natural Language Understanding and Generation
Dong, L., Yang, N., Wang, W., Wei, F., Liu, X., Wang, Y., Gao, J., Zhou, M., & Hon, H.-W. (2019) · arXiv · doi:10.48550/arxiv.1905.03197 · Cited by 1,225
“Waveglow: a flow-based generative network for speech synthesis,”
Prenger, R., Valle, R., & Catanzaro, B. (2019) · ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) · doi:10.1109/icassp.2019.8683143 · Cited by 1,004
Texygen: A Benchmarking Platform for Text Generation Models
Zhu, Y., Lu, S., Zheng, L., Guo, J., Zhang, W., Wang, J., & Yu, Y. (2018) · The 41st International ACM SIGIR Conference on Research & Development in Information Retrieval · doi:10.1145/3209978.3210080 · Cited by 408
Towards Better Decoding and Language Model Integration in Sequence to Sequence Models
Chorowski, J., & Jaitly, N. (2017) · Interspeech 2017 · doi:10.21437/interspeech.2017-343 · Cited by 358 · cited 1×
“We compute either a PER (without fusing additional word/character-based LM (Chorowski and Jaitly, 2016) or imposing lexical constraints), or a CER after the LM (we use a CER instead of a WER to give partial credit).”
Research works printed but not linked: 58
- Baevski, 2019, Effectiveness of self-supervised pre-training for speech recognition, CoRR
- Baevski, 2021, Unsupervised speech recognition, arXiv preprint arXiv:2012.15454 · arXiv
- Baevski, 2020, vq-wav2vec: Self-supervised learning of discrete speech representations, International Conference on Learning Representations (ICLR)
- Baevski, 2020, wav2vec 2.0: A framework for self-supervised learning of speech representations, 12449
- Brown, 2020, Language models are few-shot learners
- Chen, 2020, Unsupervised acoustic unit representation learning for voice conversion using WaveNet auto- encoders, 4866, 10.21437/Interspeech.2020-1785 · DOI
- Chi, 2021, Audio ALBERT: A lite BERT for self-supervised learning of audio representation, 344, 10.1109/SLT48900.2021.9383575 · DOI · Paper page
- Chung, 2020, Improved speech representations with multi-target autoregressive predictive coding, 2353, 10.18653/v1/2020.acl-main.213 · DOI · Paper page
- Chung, 2019, An unsupervised autoregressive model for speech representation learning, 146, 10.21437/Interspeech.2019-1473 · DOI · Paper page
- Dunbar, 2019, The Zero Resource Speech Challenge 2019: TTS without T, 1088, 10.21437/Interspeech.2019-2904 · DOI · Paper page
- Dunbar, 2020, The Zero Resource Speech Challenge 2020: Discovering discrete subword and word units, 4831, 10.21437/Interspeech.2020-2743 · DOI
- Dupoux, 2018, Cognitive science in the era of artificial intelligence: A roadmap for reverse-engineering the infant language- learner, Cognition, 173, 43, 10.1016/j.cognition.2017.11.008 · DOI · Paper page
- Ebbers, 2017, Hidden Markov Model variational autoencoder for acoustic unit discovery, 488, 10.21437/Interspeech.2017-1160 · DOI · Paper page
- Eloff, 2019, Unsupervised acoustic unit discovery for speech synthesis using discrete latent-variable neural networks, 1103, 10.21437/Interspeech.2019-1518 · DOI · Paper page
- Feng, 2019, Combining adversarial training and disentangled speech representation for robust zero- resource subword modeling, 1093, 10.21437/Interspeech.2019-1337 · DOI
- Glarner, 2018, Full Bayesian Hidden Markov Model variational autoencoder for acoustic unit discovery, 2688, 10.21437/Interspeech.2018-2148 · DOI
- Hsu, 2020, Text-free image- to-speech synthesis using learned segmental units, arXiv preprint arXiv:2012.15454 · arXiv
- Hsu, 2021, HuBERT: How much can a bad teacher benefit ASR pre-training?, 6533, 10.1109/ICASSP39728.2021.9414460 · DOI · Paper page
- Hsu, 2017, Learning latent representations for speech generation and transformation, 1273, 10.21437/Interspeech.2017-349 · DOI · Paper page
- Hsu, 2017, Unsupervised learning of disentangled and interpretable representations from sequential data, 1878
- Jin, 2018, FFTNet: A real-time speaker-dependent neural vocoder, 2251, 10.1109/ICASSP.2018.8462431 · DOI
- Kahn, 2020, Libri-light: A benchmark for ASR with limited or no supervision, 7669, 10.1109/ICASSP40776.2020.9052942 · DOI · Paper page
- Kharitonov, 2021, Data augmenting contrastive learning of speech representations in the time domain, arXiv preprint arXiv:2007.00991, 215, 10.1109/SLT48900.2021.9383605 · DOI
- Khurana, 2019, A factorial deep Markov Model for unsupervised disentangled representation learning from speech, 6540
- Khurana, 2020, A convolutional deep Markov model for unsupervised speech representation learning, 3790, 10.21437/Interspeech.2020-3084 · DOI
- Kong, 2020, HiFi-GAN: Generative adversarial networks for efficient and high fidelity speech synthesis, 17022
- Kumar, 2019, MelGAN: Generative adversarial networks for conditional waveform synthesis, 14910
- Lai, 2021, Semi- supervised spoken language understanding via self-supervised speech and language model pretraining, 7468, 10.1109/ICASSP39728.2021.9414922 · DOI
- Lee, 2012, A nonparametric Bayesian approach to acoustic model discovery, 40
- Ling, 2020, DeCoAR 2.0: Deep contextualized acoustic representations with vector quantization, arXiv preprint arXiv:2012.06659 · arXiv
- Ling, 2020, Deep contextualized acoustic representations for semi-supervised speech recognition, 6429
- Liu, 2020, Mockingjay: Unsupervised speech representation learning with deep bidirectional transformer encoders, 6419, 10.1109/ICASSP40776.2020.9054458 · DOI · Paper page
- Liu, 2020, Non-autoregressive predictive coding for learning speech representations from local dependencies, arXiv preprint arXiv:2011.00406 · arXiv
- Liu, 2019, Unsupervised end-to-end learning of discrete linguistic units for voice conversion, 1108, 10.21437/Interspeech.2019-2048 · DOI
- Morita, 2020, Exploring TTS without T using biologically/psychologically motivated neural network modules (ZeroSpeech 2020), 4856, 10.21437/Interspeech.2020-3127 · DOI
- Nayak, 2019, Virtual phone discovery for speech synthesis without text, 10.1109/GlobalSIP45357.2019.8969412 · DOI
- Nguyen, 2020, The Zero Resource Speech Benchmark 2021: Metrics and baselines for unsupervised spoken language modeling
- Niekerk, 2020, Vector-quantized neural networks for acoustic unit discovery in the ZeroSpeech 2020 Challenge, 4836, 10.21437/Interspeech.2020-1693 · DOI · Paper page
- Ondel, 2016, Variational inference for acoustic unit discovery, Procedia Computer Science, 81, 80, 10.1016/j.procs.2016.04.033 · DOI · Paper page
- Ott, 2019, Fairseq: A fast, extensible toolkit for sequence modeling, 48
- Pascual, 2019, Learning problem-agnostic speech representations from multiple self-supervised tasks, 161, 10.21437/Interspeech.2019-2605 · DOI · Paper page
- Polyak, 2021, Speech resynthesis from discrete disentangled self-supervised representations, 10.21437/Interspeech.2021-475 · DOI · Paper page
- Polyak, 2021, High fidelity speech regeneration with application to speech enhancement, 7143, 10.1109/ICASSP39728.2021.9414853 · DOI
- Polyak, 2020, Unsupervised cross-domain singing voice conversion, 801, 10.21437/Interspeech.2020-1862 · DOI
- Polyak, 2020, TTS skins: Speaker conversion via ASR, 786, 10.21437/Interspeech.2020-1416 · DOI · Paper page
- Radford, 2019, Language models are unsupervised multitask learners, OpenAI blog, 1, 9
- Ravanelli, 2020, Multi-task self-supervised learning for robust speech recognition, 6989, 10.1109/ICASSP40776.2020.9053569 · DOI · Paper page
- Ribeiro, 2011, CROWDMOS: An approach for crowdsourcing mean opinion score studies, 2416, 10.1109/ICASSP.2011.5946971 · DOI · Paper page
- Rivière, 2020, Towards unsupervised learning of speech features in the wild, 156, 10.1109/SLT48900.2021.9383461 · DOI
- Schatz, 2021, Early phonetic learning without phonetic categories: Insights from large-scale simulations on realistic input, Proceedings of the National Academy of Sciences, 118, 10.1073/pnas.2001844118 · DOI
- Shen, 2018, Natural TTS synthesis by conditioning WaveNet on MEL spectrogram predictions, 4779, 10.1109/ICASSP.2018.8461368 · DOI · Paper page
- Tjandra, 2020, Transformer VQ-VAE for unsupervised unit discovery and speech synthesis: ZeroSpeech 2020 Challenge, 4851, 10.21437/Interspeech.2020-3033 · DOI
- Tjandra, 2019, VQVAE unsupervised unit discovery and multi-scale Code2Spec inverter for Zerospeech Challenge 2019, 1118, 10.21437/Interspeech.2019-3232 · DOI · Paper page
- Tobing, 2020, Cyclic spectral modeling for unsupervised unit discovery into voice conversion with excitation and waveform modeling, 4861, 10.21437/Interspeech.2020-2559 · DOI
- Vaswani, 2017, Attention is all you need, 5998
- Versteegh, 2016, The Zero Resource Speech Challenge 2015: Proposed approaches and results, Procedia Computer Science, 81, 67, 10.1016/j.procs.2016.04.031 · doi:10.1016/j.procs.2016.04.031
- Wang, 2020, Unsupervised pre-training of bidirectional speech encoders via masked reconstruction, 6889, 10.1109/ICASSP40776.2020.9053541 · DOI · Paper page
- Anne, 2020, Self-supervised representations improve end-to-end speech translation, 1491, 10.21437/Interspeech.2020-3094 · DOI
Cited by
A high-performance neuroprosthesis for speech decoding and avatar control
Metzger, S. L., Littlejohn, K. T., Silva, A. B., Moses, D. A., Seaton, M. P., Wang, R., Dougherty, M. E., Liu, J. R., Wu, P., Berger, M. A., et al. (2023) · Nature · doi:10.1038/s41586-023-06443-4 · Cited by 522
A review of deep learning techniques for speech processing
Mehrish, A., Majumder, N., Bharadwaj, R., Mihalcea, R., & Poria, S. (2023) · Information Fusion · doi:10.1016/j.inffus.2023.101869 · Cited by 389
Self-supervised speech representation learning: A review
Mohamed, A., Lee, H.-Y., Borgholt, L., Havtorn, J. D., Edin, J., Igel, C., Kirchhoff, K., Li, S.-W., Livescu, K., Maaløe, L., et al. (2022) · IEEE Journal of Selected Topics in Signal Processing · doi:10.1109/jstsp.2022.3207050 · Cited by 372
“The generative spoken language model (GSLM) [78] utilizes discrete representations from wav2vec 2.0, HuBERT, and CPC algorithms as inputs to an autoregressive language model trained by using the cross-entropy function to maximize the probability of predicting the next discrete speech token.”
BackgroundMethodInfluentialSelf-citationSpeechGPT: Empowering Large Language Models with Intrinsic Cross-Modal Conversational Abilities
Zhang, D., Li, S., Zhang, X., Zhan, J., Wang, P., Zhou, Y., & Qiu, X. (2023) · arXiv · doi:10.48550/arxiv.2305.11000 · Cited by 189
“…spoken language models involves encoding the speech signal into a discrete representation (Baevski et al., 2020; Hsu et al., 2021; Zhang et al., 2023a) and modeling it with language models (Lakhotia et al., 2021; Borsos et al., 2022; Zhang et al., 2023d; Wang et al., 2023; Zhang et al., 2023c).”
MethodTheory Is All You Need: AI, Human Cognition, and Causal Reasoning
(2024) · doi:10.1287/stsc.2024.0189 · Cited by 151
Metadata
- Authors
- Lakhotia, K., Kharitonov, E., Hsu, W.-N., Adi, Y., Polyak, A., Bolte, B., Nguyen, T.-A., Copet, J., Baevski, A., Mohamed, A., & Emmanuel Dupoux, A.
- Venue
- Transactions of the Association for Computational Linguistics
- Publisher
- MIT Press
- Published
- 2021
- Volume
- 9
- Pages
- 1336-1354
- ISSN
- 2307-387X (ISSN-L)
- OpenAlex
- W3129009457
- Cited by
- 75
- References
- 16
- FWCI
- 4.41
- Type
- journal article
- Language
- en
- Open access
- diamond
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.