---
title: "Towards a Unified Benchmark for Arabic — Kheir 2025"
description: "Arabic pronunciation training lacks shared, open evaluation resources."
authors: ["Yassine El Kheir", "Omnia Ibrahim", "Amit Meghanani", "Nada Almarwani", "Hawau Olamide Toyin", "Sadeen Alharbi", "Modar Alfadly", "Lamya Alkanhal", "Ibrahim Selim", "Shehab Elbatal", "Salima Mdhaffar", "Thomas Hain", "Yasser Hifny", "Mostafa Shahin", "Ahmed Ali"]
year: 2025
venue: "arXiv:2506.07722"
source: "https://arxiv.org/abs/2506.07722"
doi: "10.48550/arxiv.2506.07722"
arxiv: "2506.07722"
licence: "CC BY 4.0 (arXiv)"
licence_url: "http://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/kheir-2025-towards-unified-benchmark-arabic-pronunciation/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/kheir-2025-towards-unified-benchmark-arabic-pronunciation/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2025](https://md.research.elamur.ai/free/en/papers/2025/page/4/) / Towards a Unified Benchmark for Arabic Pronunciation Assessment: Quranic Recitation as Case Study

# Towards a Unified Benchmark for Arabic Pronunciation Assessment: Quranic Recitation as Case Study

**Authors:** [Yassine El Kheir](https://md.research.elamur.ai/free/en/authors/yassine-el-kheir/), [Omnia Ibrahim](https://md.research.elamur.ai/free/en/authors/omnia-ibrahim/), [Amit Meghanani](https://md.research.elamur.ai/free/en/authors/amit-meghanani/), [Nada Almarwani](https://md.research.elamur.ai/free/en/authors/nada-almarwani/), [Hawau Olamide Toyin](https://md.research.elamur.ai/free/en/authors/hawau-olamide-toyin/), Sadeen Alharbi, Modar Alfadly, Lamya Alkanhal, Ibrahim Selim, Shehab Elbatal, [Salima Mdhaffar](https://md.research.elamur.ai/free/en/authors/salima-mdhaffar/), [Thomas Hain](https://md.research.elamur.ai/free/en/authors/thomas-hain/) et al. (15 authors) · **Year:** 2025 · **Venue:** arXiv:2506.07722 · **Licence:** [CC BY 4.0 (arXiv)](http://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://arxiv.org/abs/2506.07722) · [PDF](https://arxiv.org/pdf/2506.07722) · [DOI 10.48550/arxiv.2506.07722](https://doi.org/10.48550/arxiv.2506.07722) · [arXiv 2506.07722](https://arxiv.org/abs/2506.07722) · [OpenAlex](https://openalex.org/W4415433250)

**Citations:** 3 citations (openalex-snapshot · 2026-10-14) · [10](#cited-by) in our library, of the 140884 papers whose bibliographies we read

## Summary
Arabic pronunciation training lacks shared, open evaluation resources. Mispronunciation detection for Modern Standard Arabic has had no common test set, and labelled error speech is scarce. The authors build an end-to-end benchmark around Qur'anic verses read in MSA style, without Tajweed. The pipeline has four parts:
- A training corpus drawn from Common Voice, with transcripts auto-vowelized and Qur'anic samples added.
- A synthetic corpus produced by TTS. Some of its inputs carry deliberately injected errors chosen from a confusion matrix.
- A 68-unit phoneme inventory adapted from an MSA phonetizer built for speech synthesis.
- A natural-speech test set, QuranMB.v1, recorded by speakers who produced prescribed errors on purpose.

Baselines follow the SUPERB recipe. A frozen SSL encoder (Wav2vec2, HuBERT, WavLM or mHuBERT) feeds a Bi-LSTM head trained with CTC to output phonemes. Detection quality stays low: the reported F1 reaches at most about 30%, and precision can drop to roughly 13%. This shows the task is far from solved. The authors also report that their synthetic data with simulated errors performs competitively with real collected speech. All data and models are released on Hugging Face under IqraEval.

## Key points
- **CMV-Ar** is an 82.37 h subset of Common Voice 12.0 (79 h train, 3.33 h dev), vowelized with an in-house tool and supplemented with Qur'anic recitations.
- **TTS data** comes from seven single-speaker voices (5 male, 2 female). They produce 26 h of correct speech plus 26 h containing injected errors, for 52 h in total. Each erroneous utterance has four characters or diacritics altered.
- **Combined training** joins real and synthetic data into 131 h. Evaluation uses the real dev set.
- **QuranMB.v1** covers 98 verses read by 18 native speakers (14 female, 4 male), about 2.2 h. A custom recording tool marked which errors each speaker should produce.
- **Phoneme set** has 68 units. Geminates are written as doubled symbols, such as /bb/. Emphatic-context vowel variants are merged because they do not change meaning.

## Interesting ideas
- Perturbing a clean canonical input with a confusion matrix before generation yields error data whose labels are exactly known, with no manual annotation needed. — For Mova (learn): Inject controlled errors into tokens or agreement in silver sentences to build negative controls, and use them to train self-induced repair rules with exact gold diffs.
- Merging distinctions that are mere realizations, and do not carry meaning, keeps the label inventory focused on what matters for the downstream judgment. — For Mova (tokens): When designing tag and lemma inventories, collapse surface variants that never change the frame or meaning, so tagger errors on them do not count against the system.
- The test set was recorded by people told exactly which deliberate error to make, so the "wrong" answers are pre-registered rather than inferred afterwards. — For Mova (answer): Build test items where a known false premise or misleading detail is planted on purpose, then check whether the reader flags exactly that spot.
- Detection works by recognizing what was actually produced and aligning it against the expected sequence, so every flagged error points to a concrete position. — For Mova (back): In the text → graph → text compressor, align the regenerated text with the original and report the differing spans as explainable loss diagnostics.
- The SUPERB-style setup freezes the encoder, takes a weighted sum over its layers and trains only a light head, which allows a fair comparison of encoders. — For Mova (tree): Probe candidate transformer encoders for gparse with frozen weights and a fixed small head, so treebank LAS differences reflect the encoder alone.

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

Fields: [Linguistics & languages](https://md.research.elamur.ai/free/en/shelves/linguistics/)

## Cited by

Cited by

10 papers of our library cite it (of the 140884 whose bibliographies we read); 3 of them have a page here, 7 not yet.

- [Harf-Speech: A Clinically Aligned Framework for Arabic Phoneme-Level Speech Assessment](https://md.research.elamur.ai/free/en/papers/azad-2026-harf-speech-clinically-aligned-framework/)
- [Enhancing Quranic Learning: A Multimodal Deep Learning Approach for Arabic Phoneme Recognition](https://md.research.elamur.ai/free/en/papers/kucukmanisa-2025-enhancing-quranic-learning-multimodal-deep/)
- [A Fusion-Aware Two-Stage Framework for Mispronunciation Detection and Diagnosis in Low-Resource Modern Standard Arabic](https://md.research.elamur.ai/free/en/papers/yang-2026-fusion-aware-two-stage-framework/)

## Cite

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

```bibtex
@misc{kheir2025unified,
  title = {{Towards a Unified Benchmark for Arabic Pronunciation Assessment: Quranic Recitation as Case Study}},
  author = {Kheir, Yassine El and Ibrahim, Omnia and Meghanani, Amit and Almarwani, Nada and Toyin, Hawau Olamide and Alharbi, Sadeen and Alfadly, Modar and Alkanhal, Lamya and Selim, Ibrahim and Elbatal, Shehab and Mdhaffar, Salima and Hain, Thomas and Hifny, Yasser and Shahin, Mostafa and Ali, Ahmed},
  year = {2025},
  doi = {10.48550/arxiv.2506.07722},
  eprint = {2506.07722},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2506.07722}
}
```

- [← previous TokenSkip: Controllable Chain-of-Thought Compression in LLMs](https://md.research.elamur.ai/free/en/papers/xia-2025-tokenskip-controllable-chain-thought-compression/)
- [next → Towards an Integrated Methodology of Dating Biblical Texts: The Case of the Book of Jeremiah](https://md.research.elamur.ai/free/en/papers/aclbulk-2025-naaijer-integrated-methodology-dating-biblical-texts/)
- [↑ back to papers of 2025](https://md.research.elamur.ai/free/en/papers/2025/page/4/)

## References

26 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. A. Neri, O. Mich, M. Gerosa, and D. Giuliani, “The effectiveness of computer assisted pronunciation training for foreign language learning by children,” Computer Assisted Language Learning, vol. 21, no. 5, pp. 393–408, 2008. [find](https://scholar.google.com/scholar?q=A%2E+Neri%2C+O%2E+Mich%2C+M%2E+Gerosa%2C+and+D%2E+Giuliani%2C+%E2%80%9CThe+effectiveness+of+computer+assisted+pronunciation+training+for+foreign+language+learning+by+children%2C%E2%80%9D+Computer+Assisted+Language+Learning%2C+vol%2E+21%2C+n)
2. Y. E. Kheir, H. Mubarak, A. Ali, and S. A. Chowdhury, “Beyond orthography: Automatic recovery of short vowels and dialectal sounds in arabic,” arXiv preprint arXiv:2408.02430, 2024. [↗](https://arxiv.org/abs/2408.02430) [our card](https://md.research.elamur.ai/free/en/papers/acl-2024-kheir-beyond-orthography-automatic-recovery-short/)
3. M. M. Al-Ghamdi, H. Al-Muhtasib, and M. Elshafei, “Phonetic rules in arabic script,” Journal of King Saud University-Computer and Information Sciences, vol. 16, pp. 85–115, 2004. [find](https://scholar.google.com/scholar?q=M%2E+M%2E+Al%2DGhamdi%2C+H%2E+Al%2DMuhtasib%2C+and+M%2E+Elshafei%2C+%E2%80%9CPhonetic+rules+in+arabic+script%2C%E2%80%9D+Journal+of+King+Saud+University%2DComputer+and+Information+Sciences%2C+vol%2E+16%2C+pp%2E+85%E2%80%93115%2C+2004%2E)
4. N. Alrashoudi, H. Al-Khalifa, and Y. Alotaibi, “Improving mispronunciation detection and diagnosis for non-native learners of the arabic language,” Discover Computing, vol. 28, no. 1, p. 1, 2025. [find](https://scholar.google.com/scholar?q=N%2E+Alrashoudi%2C+H%2E+Al%2DKhalifa%2C+and+Y%2E+Alotaibi%2C+%E2%80%9CImproving+mispronunciation+detection+and+diagnosis+for+non%2Dnative+learners+of+the+arabic+language%2C%E2%80%9D+Discover+Computing%2C+vol%2E+28%2C+no%2E+1%2C+p%2E+1%2C+2025%2E)
5. N. O. Balula, M. Rashwan, and S. Abdou, “Automatic speech recognition (asr) systems for learning arabic language and al-quran recitation: a review,” International Journal of Computer Science and Mobile Computing, vol. 10, no. 7, pp. 91–100, 2021. [find](https://scholar.google.com/scholar?q=N%2E+O%2E+Balula%2C+M%2E+Rashwan%2C+and+S%2E+Abdou%2C+%E2%80%9CAutomatic+speech+recognition+%28asr%29+systems+for+learning+arabic+language+and+al%2Dquran+recitation%3A+a+review%2C%E2%80%9D+International+Journal+of+Computer+Science+and+Mobil)
6. S. M. Witt and S. J. Young, “Phone-level pronunciation scoring and assessment for interactive language learning,” Speech communication, vol. 30, no. 2-3, pp. 95–108, 2000. [find](https://scholar.google.com/scholar?q=S%2E+M%2E+Witt+and+S%2E+J%2E+Young%2C+%E2%80%9CPhone%2Dlevel+pronunciation+scoring+and+assessment+for+interactive+language+learning%2C%E2%80%9D+Speech+communication%2C+vol%2E+30%2C+no%2E+2%2D3%2C+pp%2E+95%E2%80%93108%2C+2000%2E)
7. S. Kanters, C. Cucchiarini, and H. Strik, “The goodness of pronunciation algorithm: a detailed performance study,” 2009. [find](https://scholar.google.com/scholar?q=S%2E+Kanters%2C+C%2E+Cucchiarini%2C+and+H%2E+Strik%2C+%E2%80%9CThe+goodness+of+pronunciation+algorithm%3A+a+detailed+performance+study%2C%E2%80%9D+2009%2E)
8. M. Dhleima, M. C. Tourad, C. A. A. Telmoud, A. Ab-delmounaim, and M. F. Nanne, “Multitask learning for arabic dialects identification and machine translation,” in International Conference on Artificial Intelligence and its Applications in the Age of Digital Transformation. Springer, 2024, pp. 284–292. [find](https://scholar.google.com/scholar?q=M%2E+Dhleima%2C+M%2E+C%2E+Tourad%2C+C%2E+A%2E+A%2E+Telmoud%2C+A%2E+Ab%2Ddelmounaim%2C+and+M%2E+F%2E+Nanne%2C+%E2%80%9CMultitask+learning+for+arabic+dialects+identification+and+machine+translation%2C%E2%80%9D+in+International+Conference+on+Artificia)
9. A. Baevski, Y. Zhou, A. Mohamed, and M. Auli, “wav2vec 2.0: A framework for self-supervised learning of speech representations,” in Advances in Neural Information Processing Systems, H. Larochelle, M. Ranzato, R. Hadsell, M. Balcan, and H. Lin, Eds., vol. 33. Curran Associates, Inc., 2020, pp. 12 449–12 460. [find](https://scholar.google.com/scholar?q=A%2E+Baevski%2C+Y%2E+Zhou%2C+A%2E+Mohamed%2C+and+M%2E+Auli%2C+%E2%80%9Cwav2vec+2%2E0%3A+A+framework+for+self%2Dsupervised+learning+of+speech+representations%2C%E2%80%9D+in+Advances+in+Neural+Information+Processing+Systems%2C+H%2E+Larochelle%2C+M%2E) [our card](https://md.research.elamur.ai/free/en/papers/baevski-2020-wav2vec-2-0-framework-self/)
10. A. Ali, P. Bell, J. Glass, Y. Messaoui, H. Mubarak, S. Renals, and Y. Zhang, “The mgb-2 challenge: Arabic multi-dialect broadcast media recognition,” in 2016 IEEE Spoken Language Technology Workshop (SLT). IEEE, 2016, pp. 279–284. [find](https://scholar.google.com/scholar?q=A%2E+Ali%2C+P%2E+Bell%2C+J%2E+Glass%2C+Y%2E+Messaoui%2C+H%2E+Mubarak%2C+S%2E+Renals%2C+and+Y%2E+Zhang%2C+%E2%80%9CThe+mgb%2D2+challenge%3A+Arabic+multi%2Ddialect+broadcast+media+recognition%2C%E2%80%9D+in+2016+IEEE+Spoken+Language+Technology+Workshop+%28) [our card](https://md.research.elamur.ai/free/en/papers/ali-2016-mgb-2-challenge-arabic-multi/)
11. Ş. S. Çalık, A. Küçükmanisa, and Z. H. Kilimci, “A novel framework for mispronunciation detection of arabic phonemes using audio-oriented transformer models,” Applied Acoustics, vol. 215, p. 109711, 2024. [find](https://scholar.google.com/scholar?q=S%CC%A7%2E+S%2E+C%CC%A7al%C4%B1k%2C+A%2E+Ku%CC%88c%CC%A7u%CC%88kmanisa%2C+and+Z%2E+H%2E+Kilimci%2C+%E2%80%9CA+novel+framework+for+mispronunciation+detection+of+arabic+phonemes+using+audio%2Doriented+transformer+models%2C%E2%80%9D+Applied+Acoustics%2C+vol%2E+215%2C+p%2E+1097)
12. Y. S. Alsahafi and M. Asad, “Empirical study on mispronunciation detection for tajweed rules during quran recitation,” in 2024 6th ICCI. IEEE, 2024, pp. 39–45. [find](https://scholar.google.com/scholar?q=Y%2E+S%2E+Alsahafi+and+M%2E+Asad%2C+%E2%80%9CEmpirical+study+on+mispronunciation+detection+for+tajweed+rules+during+quran+recitation%2C%E2%80%9D+in+2024+6th+ICCI%2E+IEEE%2C+2024%2C+pp%2E+39%E2%80%9345%2E)
13. M. A. Rahman, I. A. A. Kassim, T. A. Rahman, and S. Z. M. Muji, “Development of automated tajweed checking system for children in learning quran,” Evolution in Electrical and Electronic Engineering, vol. 2, no. 1, pp. 165–176, 2021. [find](https://scholar.google.com/scholar?q=M%2E+A%2E+Rahman%2C+I%2E+A%2E+A%2E+Kassim%2C+T%2E+A%2E+Rahman%2C+and+S%2E+Z%2E+M%2E+Muji%2C+%E2%80%9CDevelopment+of+automated+tajweed+checking+system+for+children+in+learning+quran%2C%E2%80%9D+Evolution+in+Electrical+and+Electronic+Engineering%2C+v)
14. A. A. Harere and K. A. Jallad, “Quran recitation recognition using end-to-end deep learning,” arXiv preprint arXiv:2305.07034, 2023. [↗](https://arxiv.org/abs/2305.07034)
15. R. Ardila, M. Branson, K. Davis, M. Henretty, M. Kohler, J. Meyer, R. Morais, L. Saunders, F. M. Tyers, and G. Weber, “Common voice: A massively-multilingual speech corpus,” arXiv preprint arXiv:1912.06670, 2019. [↗](https://arxiv.org/abs/1912.06670) [our card](https://md.research.elamur.ai/free/en/papers/ardila-2019-common-voice-massively-multilingual-speech/)
16. D. Korzekwa, J. Lorenzo-Trueba, T. Drugman, and B. Kostek, “Computer-assisted pronunciation training—speech synthesis is almost all you need,” Speech Communication, vol. 142, pp. 22–33, 2022. [find](https://scholar.google.com/scholar?q=D%2E+Korzekwa%2C+J%2E+Lorenzo%2DTrueba%2C+T%2E+Drugman%2C+and+B%2E+Kostek%2C+%E2%80%9CComputer%2Dassisted+pronunciation+training%E2%80%94speech+synthesis+is+almost+all+you+need%2C%E2%80%9D+Speech+Communication%2C+vol%2E+142%2C+pp%2E+22%E2%80%9333%2C+2022%2E)
17. Y. E. Kheir, S. A. Chowdhury, A. Ali, H. Mubarak, and S. Afzal, “Speechblender: Speech augmentation framework for mispronunciation data generation,” arXiv preprint arXiv:2211.00923, 2022. [↗](https://arxiv.org/abs/2211.00923)
18. N. Halabi and M. Wald, “Phonetic inventory for an arabic speech corpus,” in LREC’16, 2016, pp. 734–738. [find](https://scholar.google.com/scholar?q=N%2E+Halabi+and+M%2E+Wald%2C+%E2%80%9CPhonetic+inventory+for+an+arabic+speech+corpus%2C%E2%80%9D+in+LREC%E2%80%9916%2C+2016%2C+pp%2E+734%E2%80%93738%2E)
19. S. wen Yang, P.-H. Chi, Y.-S. Chuang, C.-I. J. Lai, and K. L. et al., “Superb: Speech processing universal performance benchmark,” in Interspeech 2021, 2021, pp. 1194– 1198. [find](https://scholar.google.com/scholar?q=S%2E+wen+Yang%2C+P%2E%2DH%2E+Chi%2C+Y%2E%2DS%2E+Chuang%2C+C%2E%2DI%2E+J%2E+Lai%2C+and+K%2E+L%2E+et+al%2E%2C+%E2%80%9CSuperb%3A+Speech+processing+universal+performance+benchmark%2C%E2%80%9D+in+Interspeech+2021%2C+2021%2C+pp%2E+1194%E2%80%93+1198%2E)
20. A. Graves, S. Fernández, F. Gomez, and J. Schmidhuber, “Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks,” in ICML, ser. ICML ’06. NY, USA: Association for Computing Machinery, 2006, p. 369–376. [Online]. Available: https://doi.org/10.1145/1143844.1143891 [↗](https://doi.org/10.1145/1143844.1143891)
21. W. Hsu, B. Bolte, Y. Tsai, K. Lakhotia, R. Salakhutdinov, and A. Mohamed, “Hubert: Self-supervised speech representation learning by masked prediction of hidden units,” IEEE/ACM Trans. Audio, Speech and Lang. Proc., vol. 29, p. 3451–3460, oct 2021. [Online]. Available: https://doi.org/10.1109/TASLP.2021.3122291 [↗](https://doi.org/10.1109/taslp.2021.3122291) [our card](https://md.research.elamur.ai/free/en/papers/hsu-2021-hubert-self-supervised-speech-representation/)
22. S. Chen, C. Wang, Z. Chen, Y. Wu, S. Liu, Z. Chen, J. Li, N. Kanda, T. Yoshioka, X. Xiao, J. Wu, L. Zhou, S. Ren, Y. Qian, Y. Qian, J. Wu, M. Zeng, X. Yu, and F. Wei, “Wavlm: Large-scale self-supervised pre-training for full stack speech processing,” IEEE Journal of Selected Topics in Signal Processing, vol. 16, no. 6, pp. 1505–1518, July 2022. [find](https://scholar.google.com/scholar?q=S%2E+Chen%2C+C%2E+Wang%2C+Z%2E+Chen%2C+Y%2E+Wu%2C+S%2E+Liu%2C+Z%2E+Chen%2C+J%2E+Li%2C+N%2E+Kanda%2C+T%2E+Yoshioka%2C+X%2E+Xiao%2C+J%2E+Wu%2C+L%2E+Zhou%2C+S%2E+Ren%2C+Y%2E+Qian%2C+Y%2E+Qian%2C+J%2E+Wu%2C+M%2E+Zeng%2C+X%2E+Yu%2C+and+F%2E+Wei%2C+%E2%80%9CWavlm%3A+Large%2Dscale+self%2Dsupervis)
23. M. Zanon Boito, V. Iyer, N. Lagos, L. Besacier, and I. Calapodescu, “mhubert-147: A compact multilingual hubert model,” in Interspeech 2024, 2024, pp. 3939–3943. [find](https://scholar.google.com/scholar?q=M%2E+Zanon+Boito%2C+V%2E+Iyer%2C+N%2E+Lagos%2C+L%2E+Besacier%2C+and+I%2E+Calapodescu%2C+%E2%80%9Cmhubert%2D147%3A+A+compact+multilingual+hubert+model%2C%E2%80%9D+in+Interspeech+2024%2C+2024%2C+pp%2E+3939%E2%80%933943%2E)
24. S.-w. Yang, H.-J. Chang, Z. Huang, A. T. Liu, C.-I. Lai, H. Wu, J. Shi, X. Chang, H.-S. Tsai, W.-C. Huang et al., “A large-scale evaluation of speech foundation models,” IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2024. [find](https://scholar.google.com/scholar?q=S%2E%2Dw%2E+Yang%2C+H%2E%2DJ%2E+Chang%2C+Z%2E+Huang%2C+A%2E+T%2E+Liu%2C+C%2E%2DI%2E+Lai%2C+H%2E+Wu%2C+J%2E+Shi%2C+X%2E+Chang%2C+H%2E%2DS%2E+Tsai%2C+W%2E%2DC%2E+Huang+et+al%2E%2C+%E2%80%9CA+large%2Dscale+evaluation+of+speech+foundation+models%2C%E2%80%9D+IEEE%2FACM+Transactions+on+Audio)
25. W.-K. Leung, X. Liu, and H. Meng, “Cnn-rnn-ctc based end-to-end mispronunciation detection and diagnosis,” in ICASSP 2019. IEEE, 2019, pp. 8132–8136. [find](https://scholar.google.com/scholar?q=W%2E%2DK%2E+Leung%2C+X%2E+Liu%2C+and+H%2E+Meng%2C+%E2%80%9CCnn%2Drnn%2Dctc+based+end%2Dto%2Dend+mispronunciation+detection+and+diagnosis%2C%E2%80%9D+in+ICASSP+2019%2E+IEEE%2C+2019%2C+pp%2E+8132%E2%80%938136%2E)
26. K. Li, X. Qian, and H. Meng, “Mispronunciation detection and diagnosis in l2 english speech using multidistribution deep neural networks,” IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 25, no. 1, pp. 193–207, 2016. [find](https://scholar.google.com/scholar?q=K%2E+Li%2C+X%2E+Qian%2C+and+H%2E+Meng%2C+%E2%80%9CMispronunciation+detection+and+diagnosis+in+l2+english+speech+using+multidistribution+deep+neural+networks%2C%E2%80%9D+IEEE%2FACM+Transactions+on+Audio%2C+Speech%2C+and+Language+Processi)