---
title: "SpidR-Adapt — paper by Luthra et al. (2026)"
description: "Self-supervised speech models such as HuBERT and WavLM need thousands of hours of audio to learn useful linguistic units. 45 references."
authors: ["Mahi Luthra", "Jiayi Shen", "Maxime Poli", "Angelo Ortiz Tandazo", "Yosuke Higuchi", "Youssef Benchekroun", "Martin Gleize", "Charles-Eric Saint-James", "Dongyan Lin", "Phillip Rust", "Angel Villar", "Surya Parimi", "Vanessa Stark", "Rashel Moritz", "Juan Pino", "Yann LeCun", "Emmanuel Dupoux"]
year: 2026
venue: "Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)"
source: "https://aclanthology.org/2026.acl-long.1325/"
doi: "10.18653/v1/2026.acl-long.1325"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/luthra-2026-spidr-adapt-universal-speech-representation/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/luthra-2026-spidr-adapt-universal-speech-representation/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/50/) / SpidR-Adapt: A Universal Speech Representation Model for Few-Shot Adaptation

# SpidR-Adapt: A Universal Speech Representation Model for Few-Shot Adaptation

**Authors:** Mahi Luthra, Jiayi Shen, Maxime Poli, [Angelo Ortiz Tandazo](https://md.research.elamur.ai/free/en/authors/angelo-ortiz-tandazo/), Yosuke Higuchi, [Youssef Benchekroun](https://md.research.elamur.ai/free/en/authors/youssef-benchekroun/), [Martin Gleize](https://md.research.elamur.ai/free/en/authors/martin-gleize/), Charles-Eric Saint-James, Dongyan Lin, [Phillip Rust](https://md.research.elamur.ai/free/en/authors/phillip-rust/), Angel Villar, Surya Parimi, Vanessa Stark, Rashel Moritz, [Juan Pino](https://md.research.elamur.ai/free/en/authors/juan-pino/), [Yann LeCun](https://md.research.elamur.ai/free/en/authors/yann-lecun/) and [Emmanuel Dupoux](https://md.research.elamur.ai/free/en/authors/emmanuel-dupoux/) · **Year:** 2026 · **Venue:** [Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)](https://md.research.elamur.ai/free/en/venues/acl/) ([ACL 2026](https://md.research.elamur.ai/free/en/venues/acl/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.acl-long.1325/) · [PDF](https://aclanthology.org/2026.acl-long.1325.pdf) · [DOI 10.18653/v1/2026.acl-long.1325](https://doi.org/10.18653/v1/2026.acl-long.1325) · [OpenAlex](https://openalex.org/W7166802009)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
Self-supervised speech models such as HuBERT and WavLM need thousands of hours of audio to learn useful linguistic units. Infants manage this with a few hundred hours. SpidR-Adapt aims to narrow that gap, so that speech units can be learned for a new language from very little unlabeled audio.

The authors treat low-resource representation learning as meta-learning, using SpidR as the backbone. Their protocol, MAdaPT, divides source-language corpora into small chunks. Each chunk becomes an episode that simulates learning a language from scarce data, and key components are reset between episodes (active forgetting). The resulting bi-level problem is solved with FOBLO, a cheap first-order method. Its inner loop adapts with self-supervision, while its outer loop updates meta-parameters using a small labeled set. Training is stabilized by an initialization that alternates self-supervised steps with occasional phone-supervised steps.

With less than one hour of target-language audio, the model beats in-domain toplines, and it is roughly 100× more data-efficient than standard multi-task pre-training. This suggests a practical, architecture-agnostic route to speech representations for under-resourced languages.

## Key points
- **Three components:** MAdaPT (episodic bi-level pre-training with active forgetting), FOBLO (first-order outer-loop updates that avoid second-order cost), and interleaved light phone supervision for initialization.
- **Evaluation:**
  - Phonemic discriminability (ABX) on unseen test languages is reported as a mean across adaptation data amounts ranging between 10 minutes and 100 hours.
  - Spoken language modeling (sWUGGY, sBLIMP, tSC) uses an OPT-1.3B backbone on English.
- **Data efficiency:** Performance is comparable to monolingual SSL models trained on about 6,000 hours of data. The adapted model sees at most 100 hours of the target language, and it surpasses in-domain toplines with under 1 hour.
- **DiscoPhon benchmark:** After finetuning on 10 h with 256 units, MAdaPT-FOBLO scores the following, outperforming HuBERT, DinoSR and Multi-Task-PT SpidR:

  | Metric | MAdaPT-FOBLO | Multi-Task-PT SpidR | HuBERT |
  |---|---|---|---|
  | PER | 36.58 | 48.63 | 98.02 |
  | ABX | 4.10 | 4.47 | — |
  | PNMI | 71.64 | — | — |

  MAdaPT-Reptile performs on par with MAdaPT-FOBLO.
- **Limitations and release:** The method depends on meta-initialization and needs labeled source data in the outer loop. Training code and checkpoints are released at github.com/facebookresearch/spidr-adapt.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [Centre National de la Recherche Scientifique](https://md.research.elamur.ai/free/en/labs/centre-national-de-la-recherche-scientifique/) — Research institute, Paris, France · 169 works · 62,488 citations
- [École Normale Supérieure - PSL](https://md.research.elamur.ai/free/en/labs/ecole-normale-superieure-psl/) — Research institute, Paris, France · 23 works · 4,357 citations
- [École des hautes études en sciences sociales](https://md.research.elamur.ai/free/en/labs/ecole-des-hautes-etudes-en-sciences-sociales/) — Research institute, Paris, France · 15 works · 2,302 citations

📚 Shelves (5): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 17,450 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 17,296 works · [🔊 Signal Processing](https://md.research.elamur.ai/free/en/shelves/computer-science/signal-processing/) · 493 works

## Cite

Cite as: Mahi Luthra, Jiayi Shen, Maxime Poli, Angelo Ortiz Tandazo, Yosuke Higuchi, Youssef Benchekroun, Martin Gleize, Charles-Eric Saint-James, Dongyan Lin, Phillip Rust, Angel Villar, Surya Parimi, Vanessa Stark, Rashel Moritz, Juan Pino, Yann LeCun and Emmanuel Dupoux. 2026. SpidR-Adapt: A Universal Speech Representation Model for Few-Shot Adaptation. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{luthra2026spidr,
  title = {{SpidR-Adapt: A Universal Speech Representation Model for Few-Shot Adaptation}},
  author = {Mahi Luthra, Jiayi Shen, Maxime Poli, Angelo Ortiz Tandazo, Yosuke Higuchi, Youssef Benchekroun, Martin Gleize, Charles-Eric Saint-James, Dongyan Lin, Phillip Rust, Angel Villar, Surya Parimi, Vanessa Stark, Rashel Moritz, Juan Pino, Yann LeCun and Emmanuel Dupoux},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.1325},
  url = {https://doi.org/10.18653/v1/2026.acl-long.1325}
}
```

- [← previous SPIDE: Serial and Parallel Intertwined Speculative Decoding](https://md.research.elamur.ai/free/en/papers/xu-2026-spide-serial-parallel-intertwined-speculative/)
- [next → Spotlight and Shadow: Attention-Guided Dual-Anchor Introspective Decoding for MLLM Hallucination Mitigation](https://md.research.elamur.ai/free/en/papers/wu-2026-spotlight-shadow-attention-guided-dual/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/50/)

## References

45 references, 7 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Divyanshu Aggarwal, Ashutosh Sathe, and Sunayana Sitaram. 2025. Improving cross lingual transfer by pretraining with active forgetting. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 2367–2378, Suzhou, China. Association for Computational Linguistics… [find](https://scholar.google.com/scholar?q=Divyanshu+Aggarwal%2C+Ashutosh+Sathe%2C+and+Sunayana+Sitaram%2E+2025%2E+Improving+cross+lingual+transfer+by+pretraining+with+active+forgetting%2E+In+Proceedings+of+the+2025+Conference+on+Empirical+Methods+in+Na)
2. Emily Ahn and Eleanor Chodroff. 2022. VoxCommunis: A corpus for cross-linguistic phonetic analysis. In Proceedings of the Thirteenth Language Resources and Evaluation Conference, pages 5286–5294, Marseille, France. European Language Resources Association. [find](https://scholar.google.com/scholar?q=Emily+Ahn+and+Eleanor+Chodroff%2E+2022%2E+VoxCommunis%3A+A+corpus+for+cross%2Dlinguistic+phonetic+analysis%2E+In+Proceedings+of+the+Thirteenth+Language+Resources+and+Evaluation+Conference%2C+pages+5286%E2%80%935294%2C+Mars)
3. Rosana Ardila, Megan Branson, Kelly Davis, Michael Kohler, Josh Meyer, Michael Henretty, Reuben Morais, Lindsay Saunders, Francis Tyers, and Gregor Weber. 2020. [Common voice: A massively-multilingual speech corpus](https://md.research.elamur.ai/free/en/papers/ardila-2019-common-voice-massively-multilingual-speech/). In Proceedings of the Twelfth Language Resources and Evaluation Conference, pages 421…
4. Yu Chen, Tom Ko, and Jian Wang. 2021. A meta-learning approach for user-defined spoken term classification with varying classes and examples. In Proceedings of Interspeech 2021, pages 4224–4228. [find](https://scholar.google.com/scholar?q=Yu+Chen%2C+Tom+Ko%2C+and+Jian+Wang%2E+2021%2E+A+meta%2Dlearning+approach+for+user%2Ddefined+spoken+term+classification+with+varying+classes+and+examples%2E+In+Proceedings+of+Interspeech+2021%2C+pages+4224%E2%80%934228%2E)
5. Alejandrina Cristia. 2023. A systematic review suggests marked differences in the prevalence of infant-directed vocalization across groups of populations. Developmental Science, 26(1):e13265. [find](https://scholar.google.com/scholar?q=Alejandrina+Cristia%2E+2023%2E+A+systematic+review+suggests+marked+differences+in+the+prevalence+of+infant%2Ddirected+vocalization+across+groups+of+populations%2E+Developmental+Science%2C+26%281%29%3Ae13265%2E)
6. Margaret Cychosz, Anele Villanueva, and Adriana Weisleder. 2021. Efficient estimation of children’s language exposure in two bilingual communities. Journal of Speech, Language, and Hearing Research, 64(10):3843–3866. [find](https://scholar.google.com/scholar?q=Margaret+Cychosz%2C+Anele+Villanueva%2C+and+Adriana+Weisleder%2E+2021%2E+Efficient+estimation+of+children%E2%80%99s+language+exposure+in+two+bilingual+communities%2E+Journal+of+Speech%2C+Language%2C+and+Hearing+Research%2C+6)
7. Ewan Dunbar, Mathieu Bernard, Nicolas Hamilakis, Tu Anh Nguyen, Maureen De Seyssel, Patricia Rozé, Morgane Rivière, Eugene Kharitonov, and Emmanuel Dupoux. 2021. [find](https://scholar.google.com/scholar?q=Ewan+Dunbar%2C+Mathieu+Bernard%2C+Nicolas+Hamilakis%2C+Tu+Anh+Nguyen%2C+Maureen+De+Seyssel%2C+Patricia+Roz%C3%A9%2C+Morgane+Rivi%C3%A8re%2C+Eugene+Kharitonov%2C+and+Emmanuel+Dupoux%2E+2021%2E)
8. The Zero Resource Speech Challenge 2021: Spoken Language Modelling. In Interspeech 2021, pages 1574–1578. [find](https://scholar.google.com/scholar?q=The+Zero+Resource+Speech+Challenge+2021%3A+Spoken+Language+Modelling%2E+In+Interspeech+2021%2C+pages+1574%E2%80%931578%2E)
9. Can Balioglu, Alexander Erben, Martin Gleize, Artyom Kozhevnikov, Ilia Kulikov, and Julien Yao. 2023. fairseq2. [find](https://scholar.google.com/scholar?q=Can+Balioglu%2C+Alexander+Erben%2C+Martin+Gleize%2C+Artyom+Kozhevnikov%2C+Ilia+Kulikov%2C+and+Julien+Yao%2E+2023%2E+fairseq2%2E)
10. Ewan Dunbar, Xuan Nga Cao, Juan Benjumea, Julien Karadayi, Mathieu Bernard, Laurent Besacier, Xavier Anguera, and Emmanuel Dupoux. 2017. The zero resource speech challenge 2017. In 2017 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), pages 323–330. [find](https://scholar.google.com/scholar?q=Ewan+Dunbar%2C+Xuan+Nga+Cao%2C+Juan+Benjumea%2C+Julien+Karadayi%2C+Mathieu+Bernard%2C+Laurent+Besacier%2C+Xavier+Anguera%2C+and+Emmanuel+Dupoux%2E+2017%2E+The+zero+resource+speech+challenge+2017%2E+In+2017+IEEE+Automatic)
11. Elika Bergelson, Andrei Amatuni, Shannon Dailey, Sharath Koorathota, and Shaelise Tor. 2019. Day by day, hour by hour: Naturalistic language input to infants. Developmental Science, 22(1):e12715. [find](https://scholar.google.com/scholar?q=Elika+Bergelson%2C+Andrei+Amatuni%2C+Shannon+Dailey%2C+Sharath+Koorathota%2C+and+Shaelise+Tor%2E+2019%2E+Day+by+day%2C+hour+by+hour%3A+Naturalistic+language+input+to+infants%2E+Developmental+Science%2C+22%281%29%3Ae12715%2E)
12. Emmanuel Dupoux. 2018. Cognitive science in the era of artificial intelligence: A roadmap for reverse-engineering the infant language-learner. Cognition, 173:43–59. [find](https://scholar.google.com/scholar?q=Emmanuel+Dupoux%2E+2018%2E+Cognitive+science+in+the+era+of+artificial+intelligence%3A+A+roadmap+for+reverse%2Dengineering+the+infant+language%2Dlearner%2E+Cognition%2C+173%3A43%E2%80%9359%2E)
13. Elika Bergelson and Daniel Swingley. 2012. At 6–9 months, human infants know the meanings of many common nouns. Proceedings of the National Academy of Sciences, 109(9):3253–3258. [find](https://scholar.google.com/scholar?q=Elika+Bergelson+and+Daniel+Swingley%2E+2012%2E+At+6%E2%80%939+months%2C+human+infants+know+the+meanings+of+many+common+nouns%2E+Proceedings+of+the+National+Academy+of+Sciences%2C+109%289%29%3A3253%E2%80%933258%2E)
14. Peter D. Eimas, Eugene R. Siqueland, Peter Jusczyk, and James Vigorito. 1971. Speech perception in infants. Science, 171(3968):303–306. [find](https://scholar.google.com/scholar?q=Peter+D%2E+Eimas%2C+Eugene+R%2E+Siqueland%2C+Peter+Jusczyk%2C+and+James+Vigorito%2E+1971%2E+Speech+perception+in+infants%2E+Science%2C+171%283968%29%3A303%E2%80%93306%2E)
15. Zalán Borsos, Raphaël Marinier, Damien Vincent, Eugene Kharitonov, Olivier Pietquin, Matt Sharifi, Dominik Roblek, Olivier Teboul, David Grangier, Marco Tagliasacchi, and Neil Zeghidour. 2023. [AudioLM: A language modeling approach to audio generation](https://md.research.elamur.ai/free/en/papers/borsos-2022-audiolm-language-modeling-approach-audio/). IEEE/ACM Transactions on Audio, Speech, and Lang…
16. Sanyuan Chen, Chengyi Wang, Zhengyang Chen, Yu Wu, Shujie Liu, Zhuo Chen, Jinyu Li, Naoyuki Kanda, Takuya Yoshioka, Xiong Xiao, Jian Wu, Long Zhou, Shuo Ren, Yanmin Qian, Yao Qian, Jian Wu, Michael Zeng, Xiangzhan Yu, and Furu Wei. 2022. WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack… [find](https://scholar.google.com/scholar?q=Sanyuan+Chen%2C+Chengyi+Wang%2C+Zhengyang+Chen%2C+Yu+Wu%2C+Shujie+Liu%2C+Zhuo+Chen%2C+Jinyu+Li%2C+Naoyuki+Kanda%2C+Takuya+Yoshioka%2C+Xiong+Xiao%2C+Jian+Wu%2C+Long+Zhou%2C+Shuo+Ren%2C+Yanmin+Qian%2C+Yao+Qian%2C+Jian+Wu%2C+Michael+Ze)
17. Yihong Chen, Kelly Marchisio, Roberta Raileanu, David Ifeoluwa Adelani, Pontus Stenetorp, Sebastian Riedel, and Mikel Artetxe. 2023. Improving language plasticity via pretraining with active forgetting. In Advances in Neural Information Processing Systems (NeurIPS), volume 36, pages 31543–31557. [find](https://scholar.google.com/scholar?q=Yihong+Chen%2C+Kelly+Marchisio%2C+Roberta+Raileanu%2C+David+Ifeoluwa+Adelani%2C+Pontus+Stenetorp%2C+Sebastian+Riedel%2C+and+Mikel+Artetxe%2E+2023%2E+Improving+language+plasticity+via+pretraining+with+active+forgettin)
18. Chelsea Finn, Pieter Abbeel, and Sergey Levine. 2017. [Model-agnostic meta-learning for fast adaptation of deep networks](https://md.research.elamur.ai/free/en/papers/finn-2017-model-agnostic-meta-learning-fast/). In Proceedings of the 34th International Conference on Machine Learning, volume 70 of Proceedings of Machine Learning Research, pages 1126–1135. PMLR.
19. Itai Gat, Felix Kreuk, Tu Anh Nguyen, Ann Lee, Jade Copet, Gabriel Synnaeve, Emmanuel Dupoux, and Yossi Adi. 2023. Augmentation invariant discrete representation for generative spoken language modeling. In Proceedings of the 20th International Conference on Spoken Language Translation (IWSLT 2023),… [find](https://scholar.google.com/scholar?q=Itai+Gat%2C+Felix+Kreuk%2C+Tu+Anh+Nguyen%2C+Ann+Lee%2C+Jade+Copet%2C+Gabriel+Synnaeve%2C+Emmanuel+Dupoux%2C+and+Yossi+Adi%2E+2023%2E+Augmentation+invariant+discrete+representation+for+generative+spoken+language+modelin)
20. Mark Hallap, Emmanuel Dupoux, and Ewan Dunbar. 2023. Evaluating context-invariance in unsupervised speech representations. In Interspeech 2023, pages 2973–2977. [find](https://scholar.google.com/scholar?q=Mark+Hallap%2C+Emmanuel+Dupoux%2C+and+Ewan+Dunbar%2E+2023%2E+Evaluating+context%2Dinvariance+in+unsupervised+speech+representations%2E+In+Interspeech+2023%2C+pages+2973%E2%80%932977%2E)
21. Michael Hassid, Tal Remez, Tu Anh Nguyen, Itai Gat, Alexis Conneau, Felix Kreuk, Jade Copet, Alexandre Defossez, Gabriel Synnaeve, Emmanuel Dupoux, and 1 others. 2023. Textually pretrained speech language models. Advances in Neural Information Processing Systems, 36:63483–63501. [find](https://scholar.google.com/scholar?q=Michael+Hassid%2C+Tal+Remez%2C+Tu+Anh+Nguyen%2C+Itai+Gat%2C+Alexis+Conneau%2C+Felix+Kreuk%2C+Jade+Copet%2C+Alexandre+Defossez%2C+Gabriel+Synnaeve%2C+Emmanuel+Dupoux%2C+and+1+others%2E+2023%2E+Textually+pretrained+speech+lang)
22. Jui-Yang Hsu, Yuan-Jui Chen, and Hung-yi Lee. 2020. Meta-learning for end-to-end low-resource speech recognition. In Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 7844–7848. [find](https://scholar.google.com/scholar?q=Jui%2DYang+Hsu%2C+Yuan%2DJui+Chen%2C+and+Hung%2Dyi+Lee%2E+2020%2E+Meta%2Dlearning+for+end%2Dto%2Dend+low%2Dresource+speech+recognition%2E+In+Proceedings+of+the+IEEE+International+Conference+on+Acoustics%2C+Speech+and+Signal+Pr)
23. Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, and Abdelrahman Mohamed. 2021. [HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units](https://md.research.elamur.ai/free/en/papers/hsu-2021-hubert-self-supervised-speech-representation/). IEEE/ACM Transactions on Audio, Speech, and Language Processing, 29:3451– 3460.
24. J. Kahn, M. Rivière, W. Zheng, E. Kharitonov, Q. Xu, P.E. Mazaré, J. Karadayi, V. Liptchinsky, R. Collobert, C. Fuegen, T. Likhomanenko, G. Synnaeve, A. Joulin, A. Mohamed, and E. Dupoux. 2020. Librilight: A benchmark for ASR with limited or no supervision. In ICASSP 2020 - 2020 IEEE International C… [find](https://scholar.google.com/scholar?q=J%2E+Kahn%2C+M%2E+Rivi%C3%A8re%2C+W%2E+Zheng%2C+E%2E+Kharitonov%2C+Q%2E+Xu%2C+P%2EE%2E+Mazar%C3%A9%2C+J%2E+Karadayi%2C+V%2E+Liptchinsky%2C+R%2E+Collobert%2C+C%2E+Fuegen%2C+T%2E+Likhomanenko%2C+G%2E+Synnaeve%2C+A%2E+Joulin%2C+A%2E+Mohamed%2C+and+E%2E+Dupoux%2E+2020%2E+Libril)
25. Patricia K. Kuhl. 1979. Speech perception in early infancy: Perceptual constancy for spectrally dissimilar vowel categories. The Journal of the Acoustical Society of America, 66(6):1668–1679. [find](https://scholar.google.com/scholar?q=Patricia+K%2E+Kuhl%2E+1979%2E+Speech+perception+in+early+infancy%3A+Perceptual+constancy+for+spectrally+dissimilar+vowel+categories%2E+The+Journal+of+the+Acoustical+Society+of+America%2C+66%286%29%3A1668%E2%80%931679%2E)
26. Patricia K. Kuhl. 2004. Early language acquisition: cracking the speech code. Nature Reviews Neuroscience, 5:831–843. [find](https://scholar.google.com/scholar?q=Patricia+K%2E+Kuhl%2E+2004%2E+Early+language+acquisition%3A+cracking+the+speech+code%2E+Nature+Reviews+Neuroscience%2C+5%3A831%E2%80%93843%2E)
27. Kushal Lakhotia, Eugene Kharitonov, Wei-Ning Hsu, Yossi Adi, Adam Polyak, Benjamin Bolte, Tu-Anh Nguyen, Jade Copet, Alexei Baevski, Abdelrahman Mohamed, and Emmanuel Dupoux. 2021. [On Generative Spoken Language Modeling from Raw Audio](https://md.research.elamur.ai/free/en/papers/lakhotia-2021-generative-spoken-language-modeling-raw/). Transactions of the Association for Computational Linguistics, 9…
28. Alexander H. Liu, Heng-Jui Chang, Michael Auli, Wei-Ning Hsu, and Jim Glass. 2023. DinoSR: Self-Distillation and Online Clustering for Self-supervised Speech Representation Learning. Advances in Neural Information Processing Systems, 36:58346–58362. [find](https://scholar.google.com/scholar?q=Alexander+H%2E+Liu%2C+Heng%2DJui+Chang%2C+Michael+Auli%2C+Wei%2DNing+Hsu%2C+and+Jim+Glass%2E+2023%2E+DinoSR%3A+Self%2DDistillation+and+Online+Clustering+for+Self%2Dsupervised+Speech+Representation+Learning%2E+Advances+in+Neura)
29. Michael McAuliffe, Michaela Socolof, Sarah Mihuc, Michael Wagner, and Morgan Sonderegger. 2017. [find](https://scholar.google.com/scholar?q=Michael+McAuliffe%2C+Michaela+Socolof%2C+Sarah+Mihuc%2C+Michael+Wagner%2C+and+Morgan+Sonderegger%2E+2017%2E)
30. Montreal Forced Aligner: Trainable Text-Speech Alignment Using Kaldi. In Interspeech 2017, pages 498–502. [find](https://scholar.google.com/scholar?q=Montreal+Forced+Aligner%3A+Trainable+Text%2DSpeech+Alignment+Using+Kaldi%2E+In+Interspeech+2017%2C+pages+498%E2%80%93502%2E)
31. Nasrin Mostafazadeh, Michael Roth, Annie Louis, Nathanael Chambers, and James Allen. 2017. [find](https://scholar.google.com/scholar?q=Nasrin+Mostafazadeh%2C+Michael+Roth%2C+Annie+Louis%2C+Nathanael+Chambers%2C+and+James+Allen%2E+2017%2E)
32. LS-DSem 2017 shared task: The story cloze test. In Proceedings of the 2nd Workshop on Linking Models of Lexical, Sentential and Discourse-level Semantics, pages 46–51, Valencia, Spain. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=LS%2DDSem+2017+shared+task%3A+The+story+cloze+test%2E+In+Proceedings+of+the+2nd+Workshop+on+Linking+Models+of+Lexical%2C+Sentential+and+Discourse%2Dlevel+Semantics%2C+pages+46%E2%80%9351%2C+Valencia%2C+Spain%2E+Association+for)
33. Tu Anh Nguyen, Maureen de Seyssel, Patricia Rozé, Morgane Rivière, Evgeny Kharitonov, Alexei Baevski, Ewan Dunbar, and Emmanuel Dupoux. 2020. The zero resource speech benchmark 2021: Metrics and baselines for unsupervised spoken language modeling. Preprint, arXiv:2011.11588. — [arXiv:2011.11588](https://arxiv.org/abs/2011.11588) [find](https://scholar.google.com/scholar?q=Tu+Anh+Nguyen%2C+Maureen+de+Seyssel%2C+Patricia+Roz%C3%A9%2C+Morgane+Rivi%C3%A8re%2C+Evgeny+Kharitonov%2C+Alexei+Baevski%2C+Ewan+Dunbar%2C+and+Emmanuel+Dupoux%2E+2020%2E+The+zero+resource+speech+benchmark+2021%3A+Metrics+and+basel)
34. Alex Nichol, Joshua Achiam, and John Schulman. 2018. On first-order meta-learning algorithms. Preprint, arXiv:1803.02999. [find](https://scholar.google.com/scholar?q=Alex+Nichol%2C+Joshua+Achiam%2C+and+John+Schulman%2E+2018%2E+On+first%2Dorder+meta%2Dlearning+algorithms%2E+Preprint%2C+arXiv%3A1803%2E02999%2E)
35. Angelo Ortiz Tandazo, Manel Khentout, Youssef Benchekroun, Thomas Hueber, and Emmanuel Dupoux. 2025. [MauBERT: Universal phonetic inductive biases for few-shot acoustic units discovery](https://md.research.elamur.ai/free/en/papers/tandazo-2026-maubert-universal-phonetic-inductive-biases/). Preprint, arXiv:2512.19612.
36. Maxime Poli, Emmanuel Chemla, and Emmanuel Dupoux. 2024. Improving spoken language modeling with phoneme classification: A simple fine-tuning approach. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 5284–5292. [find](https://scholar.google.com/scholar?q=Maxime+Poli%2C+Emmanuel+Chemla%2C+and+Emmanuel+Dupoux%2E+2024%2E+Improving+spoken+language+modeling+with+phoneme+classification%3A+A+simple+fine%2Dtuning+approach%2E+In+Proceedings+of+the+2024+Conference+on+Empiric)
37. Maxime Poli, Emmanuel Chemla, and Emmanuel Dupoux. 2025a. fastabx: A library for efficient computation of abx discriminability. Preprint, arXiv:2505.02692. — [arXiv:2505.02692](https://arxiv.org/abs/2505.02692) [find](https://scholar.google.com/scholar?q=Maxime+Poli%2C+Emmanuel+Chemla%2C+and+Emmanuel+Dupoux%2E+2025a%2E+fastabx%3A+A+library+for+efficient+computation+of+abx+discriminability%2E+Preprint%2C+arXiv%3A2505%2E02692%2E+%E2%80%94+%5BarXiv%3A2505%2E02692%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2)
38. Maxime Poli, Manel Khentout, Angelo Ortiz Tandazo, Ewan Dunbar, Emmanuel Chemla, and Emmanuel Dupoux. 2026. DiscoPhon: Benchmarking the unsupervised discovery of phoneme inventories with discrete speech units. Preprint, arXiv:2603.18612. — [arXiv:2603.18612](https://arxiv.org/abs/2603.18612) [find](https://scholar.google.com/scholar?q=Maxime+Poli%2C+Manel+Khentout%2C+Angelo+Ortiz+Tandazo%2C+Ewan+Dunbar%2C+Emmanuel+Chemla%2C+and+Emmanuel+Dupoux%2E+2026%2E+DiscoPhon%3A+Benchmarking+the+unsupervised+discovery+of+phoneme+inventories+with+discrete+spee)
39. Maxime Poli, Mahi Luthra, Youssef Benchekroun, Yosuke Higuchi, Martin Gleize, Jiayi Shen, Robin Algayres, Yu-An Chung, Mido Assran, Juan Pino, and Emmanuel Dupoux. 2025b. SpidR: Learning fast and stable linguistic units for spoken language models without supervision. Transactions on Machine Learning… [find](https://scholar.google.com/scholar?q=Maxime+Poli%2C+Mahi+Luthra%2C+Youssef+Benchekroun%2C+Yosuke+Higuchi%2C+Martin+Gleize%2C+Jiayi+Shen%2C+Robin+Algayres%2C+Yu%2DAn+Chung%2C+Mido+Assran%2C+Juan+Pino%2C+and+Emmanuel+Dupoux%2E+2025b%2E+SpidR%3A+Learning+fast+and+stab)
40. Yuehan Qin, Yichi Zhang, Yi Nian, Xueying Ding, and Yue Zhao. 2025. MetaOOD: Automatic selection of OOD detection models. In The Thirteenth International Conference on Learning Representations. [find](https://scholar.google.com/scholar?q=Yuehan+Qin%2C+Yichi+Zhang%2C+Yi+Nian%2C+Xueying+Ding%2C+and+Yue+Zhao%2E+2025%2E+MetaOOD%3A+Automatic+selection+of+OOD+detection+models%2E+In+The+Thirteenth+International+Conference+on+Learning+Representations%2E)
41. Yun Qu, Cheems Wang, Yixiu Mao, Yiqin Lv, and Xiangyang Ji. 2025. Fast and robust: Task sampling with posterior and diversity synergies for adaptive decision-makers in randomized environments. In Proceedings of the 42nd International Conference on Machine Learning, volume 267 of Proceedings of Machi… [find](https://scholar.google.com/scholar?q=Yun+Qu%2C+Cheems+Wang%2C+Yixiu+Mao%2C+Yiqin+Lv%2C+and+Xiangyang+Ji%2E+2025%2E+Fast+and+robust%3A+Task+sampling+with+posterior+and+diversity+synergies+for+adaptive+decision%2Dmakers+in+randomized+environments%2E+In+Proc)
42. Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine Mcleavey, and Ilya Sutskever. 2023. [Robust speech recognition via large-scale weak supervision](https://md.research.elamur.ai/free/en/papers/radford-2022-robust-speech-recognition-large-scale/). In Proceedings of the 40th International Conference on Machine Learning, volume 202 of Proceedings of Machine Learning Research, pages 28492–2…
43. Okko Johannes Räsänen, Unto Kalervo Laine, and Toomas Altosaar. 2009. An improved speech segmentation quality measure: the r-value. In Interspeech 2009, pages 1851–1854. [find](https://scholar.google.com/scholar?q=Okko+Johannes+R%C3%A4s%C3%A4nen%2C+Unto+Kalervo+Laine%2C+and+Toomas+Altosaar%2E+2009%2E+An+improved+speech+segmentation+quality+measure%3A+the+r%2Dvalue%2E+In+Interspeech+2009%2C+pages+1851%E2%80%931854%2E)
44. Jenny R. Saffran, Janet F. Werker, and Lynne A. Werner. 2007. The Infant’s Auditory World: Hearing, Speech, and the Beginnings of Language, chapter 2. John Wiley & Sons, Ltd. [find](https://scholar.google.com/scholar?q=Jenny+R%2E+Saffran%2C+Janet+F%2E+Werker%2C+and+Lynne+A%2E+Werner%2E+2007%2E+The+Infant%E2%80%99s+Auditory+World%3A+Hearing%2C+Speech%2C+and+the+Beginnings+of+Language%2C+chapter+2%2E+John+Wiley+%26+Sons%2C+Ltd%2E)
45. Thomas Schatz. 2016. ABX-Discriminability Measures and Applications. Theses, Université Paris 6 (UPMC). [find](https://scholar.google.com/scholar?q=Thomas+Schatz%2E+2016%2E+ABX%2DDiscriminability+Measures+and+Applications%2E+Theses%2C+Universit%C3%A9+Paris+6+%28UPMC%29%2E)

*Updated 1 November 2026*