IQRA 2026: Interspeech Challenge on Automatic Pronunciation Assessment for Modern Standard Arabic (MSA)

Yassine El Kheir; Amit Meghanani; Mostafa Shahin; Omnia Ibrahim; Shammur Absar Chowdhury; Nada AlMarwani; Youssef Elshahawy; Ahmed Ali

cs.SDMar 31, 2026

IQRA 2026: Interspeech Challenge on Automatic Pronunciation Assessment for Modern Standard Arabic (MSA)

Authors: Yassine El Kheir, Amit Meghanani, Mostafa Shahin, Omnia Ibrahim, Shammur Absar Chowdhury, Nada AlMarwani, Youssef Elshahawy, Ahmed Ali

Organizations: DFKI & Technical University of Berlin, Germany · University of Sheffield, UK · University of New South Wales, Australia · Alexandria University, Egypt ·

Abstract

We present the findings of the second edition of the IQRA Interspeech Challenge, a challenge on automatic Mispronunciation Detection and Diagnosis (MDD) for Modern Standard Arabic (MSA). Building on the previous edition, this iteration introduces Iqra_Extra_IS26, a new dataset of authentic human mispronounced speech, complementing the existing training and evaluation resources. Submitted systems employed a diverse range of approaches, spanning CTC-based self-supervised learning models, two-stage fine-tuning strategies, and using large audio-language models. Compared to the first edition, we observe a substantial jump of 0.28 in F1-score, attributable both to novel architectures and modeling strategies proposed by participants and to the additional authentic mispronunciation data made available. These results demonstrate the growing maturity of Arabic MDD research and establish a stronger foundation for future work in Arabic pronunciation assessment.

Figures & tables

Split	Dataset	Type	Utts.	Hours	New
Train	Iqra_train	Real (MSA+Qur’an)	74k	$\sim$ 79h	–
Train	Iqra_TTS	Synthetic	55.4k	$\sim$ 52h	–
Train	Iqra_Extra_IS26	Real (mispronounced)	1,333	$\sim$ 1.5h	✓
Test	QuranMB.v2	Real (annotated)	1,643	$\sim$ 2.5h	–

Table 1: IQRA 2026 dataset summary. ✓= new this edition.

Rank	Team	F1 $\uparrow$	Prec. $\uparrow$	Recall $\uparrow$	Corr. $\uparrow$	TA $\uparrow$	FR $\downarrow$	FA $\downarrow$	CD $\uparrow$	PER $\downarrow$
1	whu-iasp	0.7201	0.7416	0.6998	0.9065	0.9825	0.0175	0.3002	0.8192	0.0365
2	UTokyo	0.7170	0.7325	0.7020	0.7269	0.9816	0.0184	0.2980	0.8142	0.0372
3	RAM	0.7157	0.6769	0.7593	0.9380	0.9739	0.0261	0.2407	0.7687	0.0405
4	SQZ_ww	0.6996	0.7004	0.6987	0.9436	0.9785	0.0215	0.3013	0.8433	0.0402
5	Najva	0.6894	0.7200	0.6613	0.9491	0.9815	0.0185	0.3387	0.8889	0.0400
6	Kalimat	0.6702	0.6666	0.6738	0.9379	0.9758	0.0242	0.3262	0.8508	0.0445

Table 2: Full leaderboard results on QuranMB.v2, sorted by F1-score.

\uparrow

: higher is better;

\downarrow

: lower is better. The organizer baseline (mHuBERT) is separated by rules. TA : True Accept rate (correct pronunciations correctly accepted); FR : False Reject rate (correct pronunciations incorrectly flagged as errors); FA : False Accept rate (mispronunciations incorrectly accepted as correct); CD : Correct Diagnosis rate (proportion of detected errors where the predicted phoneme matches the canonical form); Corr. : overall phoneme Correct Rate; PER : 1 - Accuracy. Precision and Recall are computed over correctly identified mispronunciations (True Rejects).

Automated Spoken Language Assessment

See all 15 papers

A Fusion-Aware Two-Stage Framework for Mispronunciation Detection and Diagnosis in Low-Resource Modern Standard Arabic
Jun 23, 2026Jing Yang, Shuqing Zhang, Yongyi Deng +5Domain Adaptation Speech Processing
School of Electronic Information, Wuhan University, Wuhan Hubei, 430072, China · School of Computing and Information Systems, University of Melbourne, Melbourne, Australia · Northwestern Polytechnical University +1
Beyond Acoustic Sparsity and Linguistic Bias: A Prompt-Free Paradigm for Mispronunciation Detection and Diagnosis
Apr 24, 2026Haopeng Geng, Longfei Yang, Xi Chen +3Speech Processing
Graduate School of Engineering, The University of Tokyo, Japan
NADI 2026: The Second Multidialectal Arabic Speech Processing Shared Task
Sep 22, 2026Peter Sullivan, Bashar Talafha, Ahmed Ashraf +11ASR Evaluation Spoken Language Understanding
The University of British Columbia · King Fahd University of Petroleum & Minerals · Avignon Université +4

Jun 23, 2026J/K move · Enter open · S save

A Fusion-Aware Two-Stage Framework for Mispronunciation Detection and Diagnosis in Low-Resource Modern Standard Arabic

Jing Yang, Shuqing Zhang, Yongyi Deng, Pan Li+4

School of Electronic Information, Wuhan University, Wuhan Hubei, 430072, China · School of Computing and Information Systems, University of Melbourne, Melbourne, Australia · Northwestern Polytechnical University+1

Domain Adaptation Speech Processing Automatic Speech Recognition Low-Resource Language Processing

Open paper AskPDF arXiv

Accurate phoneme recognition is pivotal for mispronunciation detection and diagnosis (MDD) in modern standard Arabic (MSA), yet remains constrained by data scarcity and the synthetic-real domain gap. This work proposes a two-stage end-to-end framework. It integrates a pre-trained encoder with causal dilated temporal convolutional networks to preserve fine-grained phonetic variations. A hierarchical two-stage strategy first learns general mappings from native/synthetic corpora, then adapts to scarce real learner data to mitigate domain shift without over-correction. Prediction stability is further enhanced via multi-checkpoint ensemble inference with N-gram rescoring. Evaluated on the QuranMB.v2 test set, our system achieves an F1-score of

0.7201

, a

63.1

% relative improvement over baseline (

0.4414

). This performance ranks at the top of the IqraEval.2 Challenge, establishing a new state-of-the-art for low-resource MSA in MDD.

Abstract

Automated Spoken Language Assessment

Explore similar work

A Fusion-Aware Two-Stage Framework for Mispronunciation Detection and Diagnosis in Low-Resource Modern Standard Arabic

Beyond Acoustic Sparsity and Linguistic Bias: A Prompt-Free Paradigm for Mispronunciation Detection and Diagnosis

NADI 2026: The Second Multidialectal Arabic Speech Processing Shared Task