---
title: "AraS2P: Arabic Speech-to-Phonemes System (2025)"
description: "The paper presents AraS2P, the system that won the Iqra'Eval 2025 shared task. 31 references, cited by 1 paper here."
authors: ["Bassam Mattar", "Mohamed Fayed", "Ayman Khalafallah"]
year: 2025
venue: "Proceedings of The Third Arabic Natural Language Processing Conference: Shared Tasks"
source: "https://aclanthology.org/2025.arabicnlp-sharedtasks.65/"
doi: "10.18653/v1/2025.arabicnlp-sharedtasks.65"
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/mattar-2025-aras2p-arabic-speech-phonemes-system/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/mattar-2025-aras2p-arabic-speech-phonemes-system/)

[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/14/) / AraS2P: Arabic Speech-to-Phonemes System

# AraS2P: Arabic Speech-to-Phonemes System

**Authors:** Bassam Mattar, Mohamed Fayed and [Ayman Khalafallah](https://md.research.elamur.ai/free/en/authors/ayman-khalafallah/) · **Year:** 2025 · **Venue:** [Proceedings of The Third Arabic Natural Language Processing Conference: Shared Tasks](https://md.research.elamur.ai/free/en/venues/arabicnlp/) ([ArabicNLP 2025](https://md.research.elamur.ai/free/en/venues/arabicnlp/2025/)) · **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/2025.arabicnlp-sharedtasks.65/) · [PDF](https://aclanthology.org/2025.arabicnlp-sharedtasks.65.pdf) · [DOI 10.18653/v1/2025.arabicnlp-sharedtasks.65](https://doi.org/10.18653/v1/2025.arabicnlp-sharedtasks.65) · [OpenAlex](https://openalex.org/W4416037785)

**Citations:** 0 citations (OpenAlex, 2026-10-14) · [1](https://md.research.elamur.ai/free/en/papers/mattar-2025-aras2p-arabic-speech-phonemes-system/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
The paper presents AraS2P, the system that won the Iqra'Eval 2025 shared task. The task is to transcribe recited Qur'anic Arabic into phonemes so that pronunciation mistakes can be detected and diagnosed (MDD). Arabic makes this hard: emphatic and pharyngeal consonants, short vowels and gemination can all change meaning.

The authors start from multilingual Wav2Vec2-BERT 2.0 and train it in two stages.
- **Stage 1:** they continue training on about 1,800 hours of public Arabic speech. The transcripts of this speech were automatically converted into phoneme strings with the MSA Phonetiser.
- **Stage 2:** they fine-tune on the official training data plus synthetic recitations. To build these, Qur'anic text is corrupted with plausible character errors and then voiced by XTTS-v2 using many speaker embeddings.

A histogram comparison showed that the phoneme distributions of the stage-1 corpora and the task data differ, especially for long vowels. Fine-tuning was needed to close that gap. The full setup with synthetic data gave the best F1 among the authors' variants. The work shows that cheap automatic labels plus targeted error simulation can compensate for scarce annotated learner speech.

## Key points
- Stage-1 corpora were Common Voice Arabic (157 h), SADA (668 h) and MASC (1,000 h), all relabeled with the MSA Phonetiser. Training ran for 800k iterations with AdamW, learning rate 1e-5 and batch size 32.
- The synthetic set is 60 hours, about 30% of the fine-tuning ("Tuning") data. The text noising applies a random, capped number of deletions, insertions and substitutions, with substitutions drawn from a per-character confusion map.
- On the leaderboard, F1 rose from 0.1923 (stage 1 only) to 0.4561 (official training data) to 0.4726 (with synthetic data).
- In the best setup, Correct Rate reached 0.8985 and Correct Diagnosis 0.6873. Adding synthetic data raised precision (0.3327 → 0.3713) but lowered recall (0.7252 → 0.6501).
- The stage-1-only model mostly failed on diacritic-related phonemes, especially shadda, which the authors link to their rarity in the stage-1 data. Because validation and test distributions seemed to differ, they submitted an intermediate checkpoint saved after 2.5 epochs.

## Interesting ideas
- A deterministic rule-based converter (here, text to phonemes) turns large unlabeled-for-the-task corpora into task-shaped training pairs, giving scale before the small gold set is used.
- Errors are simulated by corrupting a known reference with a confusion map and a capped edit count, so every error's location and type is known by construction.
- Before fine-tuning, they compared label histograms between the source and target data, and they bridged mismatched label inventories with an explicit mapping table.
- Reporting FA, FR, precision and recall separately exposed that the stage-1-only model flagged almost everything: high recall but near-useless precision.

## For Mova

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

- **I1** (learn): rule-based converters (e.g., UD-dialect converters) could relabel extra treebank text into Mova's target format as a silver pre-stage, followed by fine-tuning on gold data.
- **I2** (tokens): injecting controlled typos, wrong inflections or dropped apostrophes into clean text would give the tokenizer and lemmatizer free, labeled robustness tests and training data.
- **I3** (tree): comparing deprel/UPOS frequency histograms across treebanks and keeping an explicit, inspectable mapping table would show where dialect conversion or domain shift will hurt the parser.
- **I4** (learn): evaluations should break errors down by direction so that a degenerate "always flag" or "always accept" behavior is caught, much like a negative control.

*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

- [Georgia Institute of Technology](https://md.research.elamur.ai/free/en/universities/georgia-institute-of-technology/) — University, Atlanta, United States · 133 works · 24,497 citations
- [Alexandria University](https://md.research.elamur.ai/free/en/universities/alexandria-university/) — University, Alexandria, Egypt · 39 works · 287 citations

📚 Shelves (7): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works · [💭 Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/) · 1,552 works · [💭 Experimental and Cognitive Psychology](https://md.research.elamur.ai/free/en/shelves/psychology/experimental-and-cognitive-psychology/) · 870 works · [🔊 Signal Processing](https://md.research.elamur.ai/free/en/shelves/computer-science/signal-processing/) · 452 works

🧭 Topics (1): [Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) · 2,908 works

## Cite

Cite as: Bassam Mattar, Mohamed Fayed and Ayman Khalafallah. 2025. AraS2P: Arabic Speech-to-Phonemes System. Proceedings of The Third Arabic Natural Language Processing Conference: Shared Tasks.

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

```bibtex
@inproceedings{mattar2025aras2p,
  title = {{AraS2P: Arabic Speech-to-Phonemes System}},
  author = {Bassam Mattar, Mohamed Fayed and Ayman Khalafallah},
  booktitle = {Proceedings of The Third Arabic Natural Language Processing Conference: Shared Tasks},
  year = {2025},
  doi = {10.18653/v1/2025.arabicnlp-sharedtasks.65},
  url = {https://doi.org/10.18653/v1/2025.arabicnlp-sharedtasks.65}
}
```

- [← previous AraNLP at MAHED 2025 Shared Task: Using AraBERT for Text-based Hate and Hope Speech Classification](https://md.research.elamur.ai/free/en/papers/kassas-2025-aranlp-mahed-2025-shared-task/)
- [next → AraSim: Optimizing Arabic Dialect Translation in Children's Literature with LLMs and Similarity Scores](https://md.research.elamur.ai/free/en/papers/bouomar-2025-arasim-optimizing-arabic-dialect-translation/)
- [↑ back to papers of 2025](https://md.research.elamur.ai/free/en/papers/2025/page/14/)

## References

31 references, 12 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Sherif Mahdy Abdou and Mohsen Rashwan. 2014. A computer aided pronunciation learning system for teaching the holy quran recitation rules. In 2014 IEEE/ACS 11th International Conference on Computer Systems and Applications (AICCSA), pages 543–550. IEEE. [find](https://scholar.google.com/scholar?q=Sherif+Mahdy+Abdou+and+Mohsen+Rashwan%2E+2014%2E+A+computer+aided+pronunciation+learning+system+for+teaching+the+holy+quran+recitation+rules%2E+In+2014+IEEE%2FACS+11th+International+Conference+on+Computer+Sys)
2. Mohammad Al-Fetyani, Muhammad Al-Barham, Gheith Abandah, Adham Alsharkawi, and Maha Dawas. 2023. Masc: Massive arabic speech corpus. In IEEE-SLT. [find](https://scholar.google.com/scholar?q=Mohammad+Al%2DFetyani%2C+Muhammad+Al%2DBarham%2C+Gheith+Abandah%2C+Adham+Alsharkawi%2C+and+Maha+Dawas%2E+2023%2E+Masc%3A+Massive+arabic+speech+corpus%2E+In+IEEE%2DSLT%2E)
3. Sadeen Alharbi, Areeb Alowisheq, Zoltán Tüske, Kareem Darwish, Abdullah Alrajeh, Abdulmajeed Alrowithi, Aljawharah Bin Tamran, Asma Ibrahim, Raghad Aloraini, Raneem Alnajim, and 1 others. 2024. Sada: Saudi audio dataset for arabic. In ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 10286–10290. IEEE. [find](https://scholar.google.com/scholar?q=Sadeen+Alharbi%2C+Areeb+Alowisheq%2C+Zolt%C3%A1n+T%C3%BCske%2C+Kareem+Darwish%2C+Abdullah+Alrajeh%2C+Abdulmajeed+Alrowithi%2C+Aljawharah+Bin+Tamran%2C+Asma+Ibrahim%2C+Raghad+Aloraini%2C+Raneem+Alnajim%2C+and+1+others%2E+2024%2E+Sada%3A+)
4. Ahmed Ali, Peter Bell, James Glass, Yacine Messaoui, Hamdy Mubarak, Steve Renals, and Yifan Zhang. 2016. [The mgb-2 challenge: Arabic multi-dialect broadcast media recognition](https://md.research.elamur.ai/free/en/papers/ali-2016-mgb-2-challenge-arabic-multi/). pages 279–284.
5. Norah Alrashoudi, Hend Al-Khalifa, and Yousef Alotaibi. 2025. Improving mispronunciation detection and diagnosis for non-native learners of the arabic language. Discover Computing, 28(1):1. [↗](https://doi.org/10.1007/s10791-024-09489-8)
6. Sarah S Alrumiah and Amal A Al-Shargabi. 2023. Intelligent quran recitation recognition and verification: Research trends and open issues. Arabian Journal for Science and Engineering, 48(8):9859–9885. [↗](https://doi.org/10.1007/s13369-022-07273-8)
7. Duarte M Alves, José Pombal, Nuno M Guerreiro, Pedro H Martins, João Alves, Amin Farajian, Ben Peters, Ricardo Rei, Patrick Fernandes, Sweta Agrawal, and 1 others. 2024. [Tower: An open multilingual large language model for translation-related tasks](https://md.research.elamur.ai/free/en/papers/alves-2024-tower-open-multilingual-large-language/). arXiv preprint arXiv:2402.17733. [↗](https://arxiv.org/abs/2402.17733)
8. Rosana Ardila, Megan Branson, Kelly Davis, Michael Henretty, Michael Kohler, Josh Meyer, Reuben Morais, Lindsay Saunders, Francis M Tyers, and Gregor Weber. 2019. [Common voice: A massively-multilingual speech corpus](https://md.research.elamur.ai/free/en/papers/ardila-2019-common-voice-massively-multilingual-speech/). arXiv preprint arXiv:1912.06670. [↗](https://arxiv.org/abs/1912.06670)
9. Alexei Baevski, Steffen Schneider, and Michael Auli. 2019. vq-wav2vec: Self-supervised learning of discrete speech representations. arXiv preprint arXiv:1910.05453. [↗](https://arxiv.org/abs/1910.05453)
10. Alexei Baevski, Yuhao Zhou, Abdelrahman Mohamed, and Michael Auli. 2020. wav2vec 2.0: A framework for self-supervised learning of speech representations. Advances in neural information processing systems, 33:12449–12460. Loïc Barrault, Yu-An Chung, Mariano Coria Meglioli, David Dale, Ning Dong, Mark Duppenthaler, Paul-Ambroise Duquenne, Brian Ellis, Hady Elsahar, Justin Haaheim, and 1 others. 2023. [Seamless: Multilingual expressive and streaming speech translation](https://md.research.elamur.ai/free/en/papers/communication-2023-seamless-multilingual-expressive-streaming-speech/). arXiv preprint arXiv:2312.05187. [↗](https://arxiv.org/abs/2312.05187)
11. Patrizia Bonaventura, Daniel Herron, and Wolfgang Menzel. 2000. Phonetic rules for diagnosis of pronunciation errors. In KONVENS 2000/Sprachkommunikation, Vorträge der gemeinsamen Veranstaltung 5. Konferenz zur Verarbeitung natürlicher Sprache (KONVENS), 6. ITG-Fachtagung" Sprachkommunikation", pages 225–230. [find](https://scholar.google.com/scholar?q=Patrizia+Bonaventura%2C+Daniel+Herron%2C+and+Wolfgang+Menzel%2E+2000%2E+Phonetic+rules+for+diagnosis+of+pronunciation+errors%2E+In+KONVENS+2000%2FSprachkommunikation%2C+Vortr%C3%A4ge+der+gemeinsamen+Veranstaltung+5%2E+Kon)
12. Sükrü Selim Calık, Ayhan Kucukmanisa, and Zeynep Hilal Kilimci. 2023. [An ensemble-based framework for mispronunciation detection of arabic phonemes](https://md.research.elamur.ai/free/en/papers/calik-2023-ensemble-based-framework-mispronunciation-detection/). Applied Acoustics, 212:109593. [↗](https://doi.org/10.1016/j.apacoust.2023.109593)
13. Edresson Casanova, Kelly Davis, Eren Gölge, Görkem Göknar, Iulian Gulea, Logan Hart, Aya Aljafari, Joshua Meyer, Reuben Morais, Samuel Olayemi, and 1 others. 2024. Xtts: a massively multilingual zero-shot text-to-speech model. arXiv preprint arXiv:2406.04904. [↗](https://arxiv.org/abs/2406.04904)
14. Ting Chen, Simon Kornblith, Mohammad Norouzi, and Geoffrey Hinton. 2020. [A simple framework for contrastive learning of visual representations](https://md.research.elamur.ai/free/en/papers/chen-2020-simple-framework-contrastive-learning-visual/). In International conference on machine learning, pages 1597–1607. PmLR. [↗](https://doi.org/10.48550/arxiv.2002.05709)
15. Yu-An Chung, Yu Zhang, Wei Han, Chung-Cheng Chiu, James Qin, Ruoming Pang, and Yonghui Wu. 2021. W2v-bert: Combining contrastive learning and masked language modeling for self-supervised speech pre-training. In 2021 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), pages 244–250. IEEE. [find](https://scholar.google.com/scholar?q=Yu%2DAn+Chung%2C+Yu+Zhang%2C+Wei+Han%2C+Chung%2DCheng+Chiu%2C+James+Qin%2C+Ruoming+Pang%2C+and+Yonghui+Wu%2E+2021%2E+W2v%2Dbert%3A+Combining+contrastive+learning+and+masked+language+modeling+for+self%2Dsupervised+speech+pre%2Dtr)
16. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. [Bert: Pre-training of deep bidirectional transformers for language understanding](https://md.research.elamur.ai/free/en/papers/devlin-2019-bert-pre-training-deep-bidirectional/). In 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), pages 4171–4186. [↗](https://doi.org/10.18653/v1/n19-1423)
17. Yassine El Kheir, Amit Meghanani, Hawau Olamide Toyin, Nada Almarwani, Omnia Ibrahim, Youssef Elshahawy, Mostafa Shahin, and Ahmed Ali. 2025. [Iqra’eval: A shared task on qur’anic pronunciation assessment](https://md.research.elamur.ai/free/en/papers/kheir-2025-iqra-eval-shared-task-qur/). In Proceedings of the Third Arabic Natural Language Processing Conference. Association for Computational Linguistics.
18. Maxine Eskenazi. 2009. An overview of spoken language technology for education. Speech communication, 51(10):832–844. [↗](https://doi.org/10.1016/j.specom.2009.04.005)
19. Kazuki Fujii, Taishi Nakamura, Mengsay Loem, Hiroki Iida, Masanari Ohi, Kakeru Hattori, Hirai Shota, Sakae Mizuki, Rio Yokota, and Naoaki Okazaki. 2024. [Continual pre-training for cross-lingual llm adaptation: Enhancing japanese language capabilities](https://md.research.elamur.ai/free/en/papers/fujii-2024-continual-pre-training-cross-lingual/). arXiv preprint arXiv:2404.17790. [↗](https://arxiv.org/abs/2404.17790)
20. Alex Graves, Santiago Fernández, Faustino Gomez, and Jürgen Schmidhuber. 2006. Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks. In Proceedings of the 23rd international conference on Machine learning, pages 369–376. [find](https://scholar.google.com/scholar?q=Alex+Graves%2C+Santiago+Fern%C3%A1ndez%2C+Faustino+Gomez%2C+and+J%C3%BCrgen+Schmidhuber%2E+2006%2E+Connectionist+temporal+classification%3A+labelling+unsegmented+sequence+data+with+recurrent+neural+networks%2E+In+Proceedings)
21. Ahmad Al Harere and Khloud Al Jallad. 2023. Quran recitation recognition using end-to-end deep learning. arXiv preprint arXiv:2305.07034. [↗](https://arxiv.org/abs/2305.07034)
22. Kaiming He, Haoqi Fan, Yuxin Wu, Saining Xie, and Ross Girshick. 2020. [Momentum contrast for unsupervised visual representation learning](https://md.research.elamur.ai/free/en/papers/he-2019-momentum-contrast-unsupervised-visual-representation/). In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9729–9738.
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. [↗](https://doi.org/10.1109/taslp.2021.3122291)
24. Katikapalli Subramanyam Kalyan, Ajit Rajasekharan, and Sivanesan Sangeetha. 2021. Ammus: A survey of transformer-based pretrained models in natural language processing. arXiv preprint arXiv:2108.05542. [↗](https://arxiv.org/abs/2108.05542)
25. Yassine El Kheir, Ahmed Ali, and Shammur Absar Chowdhury. 2023. Automatic pronunciation assessment–a review. arXiv preprint arXiv:2310.13974. [↗](https://arxiv.org/abs/2310.13974)
26. Yassine El Kheir, Omnia Ibrahim, Amit Meghanani, Nada Almarwani, Hawau Olamide Toyin, Sadeen Alharbi, Modar Alfadly, Lamya Alkanhal, Ibrahim Selim, Shehab Elbatal, and 1 others. 2025. [Towards a unified benchmark for arabic pronunciation assessment: Quranic recitation as case study](https://md.research.elamur.ai/free/en/papers/kheir-2025-towards-unified-benchmark-arabic-pronunciation/). arXiv preprint arXiv:2506.07722. [↗](https://arxiv.org/abs/2506.07722)
27. Ilya Loshchilov and Frank Hutter. 2017. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101. [↗](https://arxiv.org/abs/1711.05101)
28. Linkai Peng, Yingming Gao, Binghuai Lin, Dengfeng Ke, Yanlu Xie, and Jinsong Zhang. 2022. Text-aware end-to-end mispronunciation detection and diagnosis. arXiv preprint arXiv:2206.07289. [↗](https://arxiv.org/abs/2206.07289)
29. Antoine Raux and Tatsuya Kawahara. 2002. Automatic intelligibility assessment and diagnosis of critical pronunciation errors for computer-assisted pronunciation learning. In INTERSPEECH, pages 737–740. [find](https://scholar.google.com/scholar?q=Antoine+Raux+and+Tatsuya+Kawahara%2E+2002%2E+Automatic+intelligibility+assessment+and+diagnosis+of+critical+pronunciation+errors+for+computer%2Dassisted+pronunciation+learning%2E+In+INTERSPEECH%2C+pages+737%E2%80%93740)
30. Silke M Witt and Steve J Young. 2000. Phone-level pronunciation scoring and assessment for interactive language learning. Speech communication, 30(2-3):95–108. [↗](https://doi.org/10.1016/s0167-6393(99)00044-8)
31. Da-Wei Zhou, Hai-Long Sun, Jingyi Ning, Han-Jia Ye, and De-Chuan Zhan. 2024. Continual learning with pre-trained models: A survey. arXiv preprint arXiv:2401.16386. [↗](https://arxiv.org/abs/2401.16386)

*Updated 1 November 2026*