---
title: "The Accented English Speech Recognition Challenge 2020:"
description: "AESRC2020 was a shared challenge on English spoken with different accents, which still makes… 38 references, cited by 7 papers here."
authors: ["Xian Shi", "Fan Yu", "Yizhou Lu", "Yuhao Liang", "Qiangze Feng", "Daliang Wang", "Yanmin Qian", "Lei Xie"]
year: 2021
venue: "arXiv:2102.10233"
source: "https://arxiv.org/abs/2102.10233"
doi: "10.48550/arxiv.2102.10233"
arxiv: "2102.10233"
licence: "arXiv — non-exclusive license to distribute"
licence_url: "http://arxiv.org/licenses/nonexclusive-distrib/1.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/shi-2021-accented-english-speech-recognition-challenge/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/shi-2021-accented-english-speech-recognition-challenge/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2021](https://md.research.elamur.ai/free/en/papers/2021/page/6/) / The Accented English Speech Recognition Challenge 2020: Open Datasets, Tracks, Baselines, Results and Methods

# The Accented English Speech Recognition Challenge 2020: Open Datasets, Tracks, Baselines, Results and Methods

**Authors:** [Xian Shi](https://md.research.elamur.ai/free/en/authors/xian-shi/), Fan Yu, Yizhou Lu, Yuhao Liang, Qiangze Feng, Daliang Wang, Yanmin Qian and Lei Xie · **Year:** 2021 · **Venue:** arXiv:2102.10233 · **Licence:** [arXiv — non-exclusive license to distribute](http://arxiv.org/licenses/nonexclusive-distrib/1.0/)

**Links:** [Publisher's page](https://arxiv.org/abs/2102.10233) · [PDF](https://arxiv.org/pdf/2102.10233) · [DOI 10.48550/arxiv.2102.10233](https://doi.org/10.48550/arxiv.2102.10233) · [arXiv 2102.10233](https://arxiv.org/abs/2102.10233) · [OpenAlex](https://openalex.org/W3130041558)

**Citations:** 13 citations (OpenAlex, 2026-10-14) · [7](https://md.research.elamur.ai/free/en/papers/shi-2021-accented-english-speech-recognition-challenge/cited-by/) in our library, of the 141,840 papers whose bibliographies we read

## Summary
AESRC2020 was a shared challenge on English spoken with different accents, which still makes speech recognition much less robust. The organizers released a corpus of read English collected in ten countries. Accent160 holds 160 labelled hours covering eight accents, recorded on phones in quiet rooms. A later unlabelled test set added two accents, Spanish and Canadian, that never appear in training. The challenge had two tracks: classifying the speaker's accent (track 1) and transcribing accented speech (track 2). Training data was limited to Accent160 plus Librispeech, and combining systems was not allowed. The baselines, built with ESPnet and Kaldi, showed two things. First, an accent classifier whose encoder starts from an ASR-trained encoder clearly outperforms classifiers trained from scratch. Second, end-to-end transformers beat hybrid chain models, especially when pretrained on Librispeech and fine-tuned on accented data. The paper also surveys the submitted systems. Its value is a shared, reasonably sized benchmark that tests generalization to unseen accents.

## Key points
- Accent160 covers eight accents at 20 hours each: Chinese, American, British, Korean, Japanese, Russian, Indian and Portuguese. Audio is 16 kHz mono wav with manual transcripts and speaker metadata (gender, age, region, device).
- The test sets differ by track: track 1 has 16 h (2 h per accent), while track 2 has 20 h, including 4 h of the unseen Spanish and Canadian accents.
- **Track 1 baseline accuracy on the CV set:**
  - Transformer-3L from scratch: 54.1%.
  - Transformer-12L from scratch: 47.8%, a sign of overfitting.
  - ASR-init-Transformer-12L: 76.1%.
- **Track 2 baseline average WER:**
  - The Kaldi chain model with a 3-gram LM reached 13.1–13.7%.
  - The ESPnet joint CTC/Attention transformer pretrained on Libri960 and fine-tuned on Accent160 with RNNLM fusion did best at 6.92%.
  - The same transformer trained on Accent160 alone with RNNLM fusion reached 8.63%.
- **Rules:** ROVER and any multi-system fusion were banned, LM training could use only the permitted transcripts, and a speaker list was published for carving out the CV set. Track 1 drew 25 teams.

## Interesting ideas
- Seeding a small classifier's encoder with weights learned on a richer sequence task (ASR) worked far better than training deeper models from scratch, which overfit on limited labels.
- Pretraining on a large standard corpus and then fine-tuning on the small specialised set beat simply pooling a matched amount of standard data with it.
- The test set deliberately contains categories absent from training (unseen accents), so the score measures generalization rather than memorisation of known conditions.
- The organizers released an explicit speaker list for building the dev split and warned that per-accent numbers from few speakers are not statistically reliable.
- Banning output-voting fusion such as ROVER forces comparisons between single systems, so gains come from the model itself rather than from ensembling.

## For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

- **I1** (tree): A register or genre classifier could start from the gparse encoder rather than random weights, reusing what the parser already learned about English structure.
- **I2** (tree): Train gparse on a large standard treebank first, then fine-tune on a small hand-checked set of fables and school texts, instead of mixing the two.
- **I3** (learn): Evaluation could hold out whole genres or text sources (e.g., a tale collection never used in silver data) as an out-of-domain control next to the in-domain score.
- **I4** (learn): Split data by document or author rather than by sentence, and report per-category scores only alongside how many independent sources support them.
- **I5** (learn): Report each witness parser's accuracy alone before computing the agreed silver output, so voting does not hide a weak witness.

*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

- [Shanghai Jiao Tong University](https://md.research.elamur.ai/free/en/universities/shanghai-jiao-tong-university/) — University, Shanghai, China · 63 works · 15,047 citations
- [Northwestern Polytechnical University](https://md.research.elamur.ai/free/en/universities/northwestern-polytechnical-university/) — University, Xi'an, China · 3 works · 21 citations

📚 Shelves (7): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 9,086 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 8,889 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 9,296 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 9,206 works · [💭 Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/) · 798 works · [💭 Experimental and Cognitive Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/experimental-and-cognitive-psychology/) · 463 works · [🔊 Signal Processing](https://md.research.elamur.ai/free/en/shelves/computer-science/signal-processing/) · 213 works

🧭 Topics (2): [English](https://md.research.elamur.ai/free/en/topics/english/) · 1,373 works · [🎙️ Speech & spoken language](https://md.research.elamur.ai/free/en/topics/speech/) · 233 works

## Cite

Cite as: Xian Shi, Fan Yu, Yizhou Lu, Yuhao Liang, Qiangze Feng, Daliang Wang, Yanmin Qian and Lei Xie. 2021. The Accented English Speech Recognition Challenge 2020: Open Datasets, Tracks, Baselines, Results and Methods. arXiv:2102.10233.

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

```bibtex
@misc{shi2021accented,
  title = {{The Accented English Speech Recognition Challenge 2020: Open Datasets, Tracks, Baselines, Results and Methods}},
  author = {Xian Shi, Fan Yu, Yizhou Lu, Yuhao Liang, Qiangze Feng, Daliang Wang, Yanmin Qian and Lei Xie},
  year = {2021},
  doi = {10.48550/arxiv.2102.10233},
  eprint = {2102.10233},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2102.10233}
}
```

- [← previous Correcting Arabic Soft Spelling Mistakes using BiLSTM-based Machine Learning](https://md.research.elamur.ai/free/en/papers/abandah-2021-correcting-arabic-soft-spelling-mistakes/)
- [next → Towards Robust Mispronunciation Detection and Diagnosis for L2 English Learners with Accent-Modulating Methods](https://md.research.elamur.ai/free/en/papers/jiang-2021-towards-robust-mispronunciation-detection-diagnosis/)
- [↑ back to papers of 2021](https://md.research.elamur.ai/free/en/papers/2021/page/6/)

## References

38 references, 6 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Hamid Behravan, Ville Hautamäki, Sabato Marco Siniscalchi, Tomi Kinnunen, and Chin-Hui Lee, “I-vector modeling of speech attributes for automatic foreign accent recognition,” IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 24, no. 1, pp. 29–41, 2015. [↗](https://doi.org/10.1109/taslp.2015.2489558)
2. Yusnita Ma, M. P. Paulraj, Sazali Yaacob, A. B. Shahriman, and Sathees Kumar Nataraj, “Speaker accent recognition through statistical descriptors of mel-bands spectral energy and neural network model,” 2012. [find](https://scholar.google.com/scholar?q=Yusnita+Ma%2C+M%2E+P%2E+Paulraj%2C+Sazali+Yaacob%2C+A%2E+B%2E+Shahriman%2C+and+Sathees+Kumar+Nataraj%2C+%E2%80%9CSpeaker+accent+recognition+through+statistical+descriptors+of+mel%2Dbands+spectral+energy+and+neural+network+model%2C)
3. Maryam Najafian and Martin Russell, “Automatic accent identification as an analytical tool for accent robust automatic speech recognition,” Speech Communication, vol. 122, pp. 44– 55, 2020. [↗](https://doi.org/10.1016/j.specom.2020.05.003)
4. Fadi Biadsy, Automatic dialect and accent recognition and its application to speech recognition, Ph.D. thesis, Columbia University, 2011. [find](https://scholar.google.com/scholar?q=Fadi+Biadsy%2C+Automatic+dialect+and+accent+recognition+and+its+application+to+speech+recognition%2C+Ph%2ED%2E+thesis%2C+Columbia+University%2C+2011%2E)
5. Pradeep Rangan, Sundeep Teki, and Hemant Misra, “Exploiting spectral augmentation for code-switched spoken language identification,” 2020. [↗](https://doi.org/10.48550/arxiv.2010.07130)
6. Nur Endah Safitri, Amalia Zahra, and Mirna Adriani, “Spoken language identification with phonotactics methods on minangkabau, sundanese, and javanese languages,” Procedia Computer Science, vol. 81, pp. 182–187, 2016. [find](https://scholar.google.com/scholar?q=Nur+Endah+Safitri%2C+Amalia+Zahra%2C+and+Mirna+Adriani%2C+%E2%80%9CSpoken+language+identification+with+phonotactics+methods+on+minangkabau%2C+sundanese%2C+and+javanese+languages%2C%E2%80%9D+Procedia+Computer+Science%2C+vol%2E+81%2C+pp)
7. Chithra Madhu, Anu George, and Leena Mary, “Automatic language identification for seven indian languages using higher level features,” in IEEE International Conference on Signal Processing, 2017. [find](https://scholar.google.com/scholar?q=Chithra+Madhu%2C+Anu+George%2C+and+Leena+Mary%2C+%E2%80%9CAutomatic+language+identification+for+seven+indian+languages+using+higher+level+features%2C%E2%80%9D+in+IEEE+International+Conference+on+Signal+Processing%2C+2017%2E)
8. Koji Okabe, Takafumi Koshinaka, and Koichi Shinoda, “Attentive statistics pooling for deep speaker embedding,” arXiv preprint arXiv:1803.10963, 2018. [↗](https://arxiv.org/abs/1803.10963)
9. Suwon Shon, Hao Tang, and James Glass, “Frame-level speaker embeddings for text-independent speaker recognition and analysis of end-to-end model,” in 2018 IEEE Spoken Language Technology Workshop (SLT). IEEE, 2018, pp. 1007– 1013. [find](https://scholar.google.com/scholar?q=Suwon+Shon%2C+Hao+Tang%2C+and+James+Glass%2C+%E2%80%9CFrame%2Dlevel+speaker+embeddings+for+text%2Dindependent+speaker+recognition+and+analysis+of+end%2Dto%2Dend+model%2C%E2%80%9D+in+2018+IEEE+Spoken+Language+Technology+Workshop+%28SLT)
10. Weidi Xie, Arsha Nagrani, Joon Son Chung, and Andrew Zisserman, “Utterance-level aggregation for speaker recognition in the wild,” in ICASSP 2019. IEEE, 2019, pp. 5791–5795. [find](https://scholar.google.com/scholar?q=Weidi+Xie%2C+Arsha+Nagrani%2C+Joon+Son+Chung%2C+and+Andrew+Zisserman%2C+%E2%80%9CUtterance%2Dlevel+aggregation+for+speaker+recognition+in+the+wild%2C%E2%80%9D+in+ICASSP+2019%2E+IEEE%2C+2019%2C+pp%2E+5791%E2%80%935795%2E)
11. Arsha Nagrani, Joon Son Chung, Weidi Xie, and Andrew Zisserman, “Voxceleb: Large-scale speaker verification in the wild,” Computer Speech & Language, vol. 60, pp. 101027, 2020. [↗](https://doi.org/10.1016/j.csl.2019.101027)
12. Carlos Teixeira, Isabel Trancoso, and António Serralheiro, “Accent identification,” in Proceeding of Fourth International Conference on Spoken Language Processing. ICSLP’96. IEEE, 1996, vol. 3, pp. 1784–1787. [find](https://scholar.google.com/scholar?q=Carlos+Teixeira%2C+Isabel+Trancoso%2C+and+Anto%CC%81nio+Serralheiro%2C+%E2%80%9CAccent+identification%2C%E2%80%9D+in+Proceeding+of+Fourth+International+Conference+on+Spoken+Language+Processing%2E+ICSLP%E2%80%9996%2E+IEEE%2C+1996%2C+vol%2E+3%2C+pp%2E+1)
13. Shamalee Deshpande, Sharat Chikkerur, and Venu Govindaraju, “Accent classification in speech,” in Fourth IEEE Workshop on Automatic Identification Advanced Technologies (AutoID’05). IEEE, 2005, pp. 139–143. [find](https://scholar.google.com/scholar?q=Shamalee+Deshpande%2C+Sharat+Chikkerur%2C+and+Venu+Govindaraju%2C+%E2%80%9CAccent+classification+in+speech%2C%E2%80%9D+in+Fourth+IEEE+Workshop+on+Automatic+Identification+Advanced+Technologies+%28AutoID%E2%80%9905%29%2E+IEEE%2C+2005%2C+pp%2E+13)
14. Asad Ahmed, Pratham Tangri, Anirban Panda, Dhruv Ramani, and Samarjit Karmakar, “Vfnet: A convolutional architecture for accent classification,” in 2019 IEEE 16th India Council International Conference (INDICON). IEEE, 2019, pp. 1–4. [find](https://scholar.google.com/scholar?q=Asad+Ahmed%2C+Pratham+Tangri%2C+Anirban+Panda%2C+Dhruv+Ramani%2C+and+Samarjit+Karmakar%2C+%E2%80%9CVfnet%3A+A+convolutional+architecture+for+accent+classification%2C%E2%80%9D+in+2019+IEEE+16th+India+Council+International+Conferenc)
15. Genta Indra Winata, Samuel Cahyawijaya, Zihan Liu, Zhaojiang Lin, Andrea Madotto, Peng Xu, and Pascale Fung, “Learning fast adaptation on cross-accented speech recognition,” arXiv preprint arXiv:2003.01901, 2020. [↗](https://arxiv.org/abs/2003.01901)
16. Zhong Meng, Hu Hu, Jinyu Li, Changliang Liu, Yan Huang, Yifan Gong, and Chin-Hui Lee, “L-vector: Neural label embedding for domain adaptation,” in ICASSP 2020. IEEE, 2020, pp. 7389–7393. [↗](https://doi.org/10.48550/arxiv.2004.13480)
17. Thibault Viglino, Petr Motlicek, and Milos Cernak, “End-to-end accented speech recognition.,” in INTERSPEECH, 2019, pp. 2140–2144. [find](https://scholar.google.com/scholar?q=Thibault+Viglino%2C+Petr+Motlicek%2C+and+Milos+Cernak%2C+%E2%80%9CEnd%2Dto%2Dend+accented+speech+recognition%2E%2C%E2%80%9D+in+INTERSPEECH%2C+2019%2C+pp%2E+2140%E2%80%932144%2E)
18. Bo Li and Khe Chai Sim, “Comparison of discriminative input and output transformations for speaker adaptation in the hybrid nn/hmm systems,” in Eleventh Annual Conference of the International Speech Communication Association, 2010. [find](https://scholar.google.com/scholar?q=Bo+Li+and+Khe+Chai+Sim%2C+%E2%80%9CComparison+of+discriminative+input+and+output+transformations+for+speaker+adaptation+in+the+hybrid+nn%2Fhmm+systems%2C%E2%80%9D+in+Eleventh+Annual+Conference+of+the+International+Speech+C)
19. Han Zhu, Li Wang, Pengyuan Zhang, and Yonghong Yan, “Multi-accent adaptation based on gate mechanism,” arXiv preprint arXiv:2011.02774, 2020. [↗](https://arxiv.org/abs/2011.02774)
20. Hank Liao, “Speaker adaptation of context dependent deep neural networks,” in ICASSP 2013. IEEE, 2013, pp. 7947– 7951. [find](https://scholar.google.com/scholar?q=Hank+Liao%2C+%E2%80%9CSpeaker+adaptation+of+context+dependent+deep+neural+networks%2C%E2%80%9D+in+ICASSP+2013%2E+IEEE%2C+2013%2C+pp%2E+7947%E2%80%93+7951%2E)
21. Tian Tan, Yanmin Qian, Maofan Yin, Yimeng Zhuang, and Kai Yu, “Cluster adaptive training for deep neural network,” in ICASSP 2015. IEEE, 2015, pp. 4325–4329. [find](https://scholar.google.com/scholar?q=Tian+Tan%2C+Yanmin+Qian%2C+Maofan+Yin%2C+Yimeng+Zhuang%2C+and+Kai+Yu%2C+%E2%80%9CCluster+adaptive+training+for+deep+neural+network%2C%E2%80%9D+in+ICASSP+2015%2E+IEEE%2C+2015%2C+pp%2E+4325%E2%80%934329%2E)
22. Sining Sun, Ching-Feng Yeh, Mei-Yuh Hwang, Mari Ostendorf, and Lei Xie, “Domain adversarial training for accented speech recognition,” CoRR, vol. abs/1806.02786, 2018. [↗](https://arxiv.org/abs/1806.02786)
23. Yi Chen Chen, Zhaojun Yang, Ching Feng Yeh, Mahaveer Jain, and Michael L. Seltzer, “Aipnet: Generative adversarial pre-training of accent-invariant networks for end-to-end speech recognition,” in ICASSP 2020, 2020. [↗](https://doi.org/10.48550/arxiv.1911.11935)
24. Sanghyun Yoo, Inchul Song, and Yoshua Bengio, “A highly adaptive acoustic model for accurate multi-dialect speech recognition,” in ICASSP 2019, 2019. [find](https://scholar.google.com/scholar?q=Sanghyun+Yoo%2C+Inchul+Song%2C+and+Yoshua+Bengio%2C+%E2%80%9CA+highly+adaptive+acoustic+model+for+accurate+multi%2Ddialect+speech+recognition%2C%E2%80%9D+in+ICASSP+2019%2C+2019%2E)
25. M. Chen, Zhanlei Yang, Jizhong Liang, Yanpeng Li, and Wenju Liu, “Improving deep neural networks based multi-accent mandarin speech recognition using i-vectors and accent-specific top layer,” in INTERSPEECH, 2015. [find](https://scholar.google.com/scholar?q=M%2E+Chen%2C+Zhanlei+Yang%2C+Jizhong+Liang%2C+Yanpeng+Li%2C+and+Wenju+Liu%2C+%E2%80%9CImproving+deep+neural+networks+based+multi%2Daccent+mandarin+speech+recognition+using+i%2Dvectors+and+accent%2Dspecific+top+layer%2C%E2%80%9D+in+INTER)
26. Isin Demirsahin, Oddur Kjartansson, Alexander Gutkin, and Clara Rivera, “[Open-source Multi-speaker Corpora of the English Accents in the British Isles](https://md.research.elamur.ai/free/en/papers/demirsahin-2020-open-source-multi-speaker-corpora/),” in Proceedings of The 12th Language Resources and Evaluation Conference (LREC), Marseille, France, May 2020, pp. 6532–6541, European Language Resources Association (ELRA).
27. Vassil Panayotov, Guoguo Chen, Daniel Povey, and Sanjeev Khudanpur, “Librispeech: an asr corpus based on public domain audio books,” in ICASSP 2015. IEEE, 2015, pp. 5206– 5210. [find](https://scholar.google.com/scholar?q=Vassil+Panayotov%2C+Guoguo+Chen%2C+Daniel+Povey%2C+and+Sanjeev+Khudanpur%2C+%E2%80%9CLibrispeech%3A+an+asr+corpus+based+on+public+domain+audio+books%2C%E2%80%9D+in+ICASSP+2015%2E+IEEE%2C+2015%2C+pp%2E+5206%E2%80%93+5210%2E)
28. Jonathan G Fiscus, “A post-processing system to yield reduced word error rates: Recognizer output voting error reduction (rover),” in 1997 IEEE Workshop on Automatic Speech Recognition and Understanding Proceedings. IEEE, 1997, pp. 347–354. [find](https://scholar.google.com/scholar?q=Jonathan+G+Fiscus%2C+%E2%80%9CA+post%2Dprocessing+system+to+yield+reduced+word+error+rates%3A+Recognizer+output+voting+error+reduction+%28rover%29%2C%E2%80%9D+in+1997+IEEE+Workshop+on+Automatic+Speech+Recognition+and+Understandi)
29. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Lukasz Kaiser, and Illia Polosukhin, “[Attention is all you need](https://md.research.elamur.ai/free/en/papers/vaswani-2017-attention-all-need/),” in Advances in neural information processing systems, 2017, pp. 5998–6008.
30. Houjun Huang, Xu Xiang, Yexin Yang, Rao Ma, and Yanmin Qian, “Aispeech-sjtu accent identification system for the accented english speech recognition challenge,” in Proc. ICASSP 2021. IEEE, 2021 (to appear). [↗](https://doi.org/10.48550/arxiv.2102.09828)
31. Myle Ott, Sergey Edunov, Alexei Baevski, Angela Fan, Sam Gross, Nathan Ng, David Grangier, and Michael Auli, “[fairseq: A fast, extensible toolkit for sequence modeling](https://md.research.elamur.ai/free/en/papers/ott-2019-fairseq-fast-extensible-toolkit-sequence/),” in Proceedings of NAACL-HLT 2019: Demonstrations, 2019. [↗](https://doi.org/10.48550/arxiv.1904.01038)
32. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Lukasz Kaiser, and Illia Polosukhin, “[Attention is all you need](https://md.research.elamur.ai/free/en/papers/vaswani-2017-attention-all-need/),” in Advances in neural information processing systems, 2017, pp. 5998–6008.
33. Anmol Gulati, James Qin, Chung-Cheng Chiu, Niki Parmar, Yu Zhang, Jiahui Yu, Wei Han, Shibo Wang, Zhengdong Zhang, Yonghui Wu, and Ruoming Pang, “[Conformer: Convolution-augmented transformer for speech recognition](https://md.research.elamur.ai/free/en/papers/gulati-2020-conformer-convolution-augmented-transformer-speech/),” 2020.
34. Zhanghao Wu, Zhijian Liu, Ji Lin, Yujun Lin, and Song Han, “Lite transformer with long-short range attention,” 2020. [↗](https://doi.org/10.48550/arxiv.2004.11886)
35. William Chan, Navdeep Jaitly, Quoc Le, and Oriol Vinyals, “Listen, attend and spell: A neural network for large vocabulary conversational speech recognition,” in ICASSP 2016. IEEE, 2016, pp. 4960–4964. [find](https://scholar.google.com/scholar?q=William+Chan%2C+Navdeep+Jaitly%2C+Quoc+Le%2C+and+Oriol+Vinyals%2C+%E2%80%9CListen%2C+attend+and+spell%3A+A+neural+network+for+large+vocabulary+conversational+speech+recognition%2C%E2%80%9D+in+ICASSP+2016%2E+IEEE%2C+2016%2C+pp%2E+4960%E2%80%934964)
36. Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, and Michael Auli, “[wav2vec 2.0: A framework for self-supervised learning of speech representations](https://md.research.elamur.ai/free/en/papers/baevski-2020-wav2vec-2-0-framework-self/),” 2020.
37. Xingchen Song, Guangsen Wang, Zhiyong Wu, Yiheng Huang, Dan Su, Dong Yu, and Helen Meng, “Speech-xlnet: Unsupervised acoustic model pretraining for self-attention networks,” 2020. [find](https://scholar.google.com/scholar?q=Xingchen+Song%2C+Guangsen+Wang%2C+Zhiyong+Wu%2C+Yiheng+Huang%2C+Dan+Su%2C+Dong+Yu%2C+and+Helen+Meng%2C+%E2%80%9CSpeech%2Dxlnet%3A+Unsupervised+acoustic+model+pretraining+for+self%2Dattention+networks%2C%E2%80%9D+2020%2E)
38. Tian Tan, Yizhou Lu, Rao Ma, Sen Zhu, Jiaqi Guo, and Yanmin Qian, “Aispeech-sjtu asr system for the accented english speech recognition challenge,” in Proc. ICASSP 2021. IEEE, 2021 (to appear). [find](https://scholar.google.com/scholar?q=Tian+Tan%2C+Yizhou+Lu%2C+Rao+Ma%2C+Sen+Zhu%2C+Jiaqi+Guo%2C+and+Yanmin+Qian%2C+%E2%80%9CAispeech%2Dsjtu+asr+system+for+the+accented+english+speech+recognition+challenge%2C%E2%80%9D+in+Proc%2E+ICASSP+2021%2E+IEEE%2C+2021+%28to+appear%29%2E)

*Updated 24 October 2026*