A Comparative Study of Pretrained Transformer Models for Quranic ASR: Speech Representations, Label Formats, and Dataset Composition

2026-06-17

AI Summary

This study addresses the high word error rates and limited coverage of existing automatic speech recognition (ASR) systems on Quranic recitation tasks. It presents the first systematic evaluation and comparison of fine-tuned self-supervised pretrained Transformer models—Wav2Vec2.0, HuBERT, and XLS-R—leveraging over 870 hours of expert and user recitation data. Through multidimensional ablation studies, the work investigates the impact of speech representations, label formats, and data composition. The results demonstrate that combining diacritic-free Arabic text with XLS-R-53 yields optimal performance, achieving word error rates (WER) of 0.08 on the EveryAyah subset and 0.11 on the EveryAyah+Tarteel combined set—substantially outperforming the Citrinet baseline (WER=0.163) by approximately five percentage points—while reducing training time from 140 to 40 hours.

Abstract

Quran Automatic Speech Recognition (ASR) aims to convert Quranic recitation into text, enabling applications such as aided memorisation tools and Quranic search engines. However, existing ASR models often exhibit high Word Error Rates (WER) on user-recited verses and lack full coverage of the Quranic corpus. This paper presents a systematic empirical study of domain-specific fine-tuning of pretrained Transformer-based models for Quranic ASR, using advanced speech feature extraction methods: Wav2Vec2.0, HuBERT, and XLS-R. These models apply self-supervised learning by masking portions of input audio and using Transformer architectures to learn context-aware speech features. The pretrained models are fine-tuned on a filtered Quranic dataset exceeding 870 hours of professional and user recitations. Through comprehensive ablation studies across feature extractors, output label formats, training strategies, and clip durations, we identify the key factors that affect transcription accuracy in this domain. Our best-performing configuration achieves a WER of 0.08 on the EveryAyah subset and 0.11 on the combined EveryAyah+Tarteel setting, representing roughly a five-percentage-point gain over the Citrinet baseline (WER = 0.163) while reducing combined-model training time from 140 hours to 40 hours. Arabic text without diacritics yields the best fine-tuning results, and Wav2Vec2-XLSR-53 provides the strongest overall representation. Future work includes improving dataset quality and developing phoneme-aware models to extract deeper speech feature representations for Tajweed-sensitive applications.

Problem

• Quranic ASR
• Word Error Rate
• speech recognition
• domain-specific fine-tuning
• Quranic corpus coverage

Innovation

• Quranic ASR
• Pretrained Transformers
• Self-supervised Learning
• Wav2Vec2-XLSR
• Word Error Rate