Improving Accuracy of Low-resource ASR using Rule-Based Character Constituency Loss (RBCCL)

Rupak Raj Ghimire, Prakash Poudyal, Bal Krishna Bal


Abstract
Modern general-purpose speech recognition systems are more robust in languages with high resources. However, achieving state-of-the-art accuracy for low-resource languages is still challenging. To deal with this challenge, one of the popular practice is fine-tuning the pre-trained model on low-resource settings. Nevertheless, pre-trained or fine-tuned model fails to capture the complex character and word constituency in the Devanagari script transcription. We proposed a complementary loss function designed to force the model to learn the character constituency of Devanagari script. Our complementary loss function, called as Rule-Based Character Constituency Loss (RBCCL), that penalizes incorrect transcriptions and updates the overall loss during the model training phase. This loss function can be combined with CTC loss or cross-entropy loss as well which are widely used in ASR training. Our experiment shows that combining the existing cross-entropy loss with new complementary loss (RBCCL) improves the Word Error Rate (WER ), reducing it from 47.1% to 23.41% which is quite promising result.
Anthology ID:
2025.chipsal-1.6
Volume:
Proceedings of the First Workshop on Challenges in Processing South Asian Languages (CHiPSAL 2025)
Month:
January
Year:
2025
Address:
Abu Dhabi, UAE
Editors:
Kengatharaiyer Sarveswaran, Ashwini Vaidya, Bal Krishna Bal, Sana Shams, Surendrabikram Thapa
Venues:
CHiPSAL | WS
SIG:
Publisher:
International Committee on Computational Linguistics
Note:
Pages:
61–70
Language:
URL:
https://aclanthology.org/2025.chipsal-1.6/
DOI:
Bibkey:
Cite (ACL):
Rupak Raj Ghimire, Prakash Poudyal, and Bal Krishna Bal. 2025. Improving Accuracy of Low-resource ASR using Rule-Based Character Constituency Loss (RBCCL). In Proceedings of the First Workshop on Challenges in Processing South Asian Languages (CHiPSAL 2025), pages 61–70, Abu Dhabi, UAE. International Committee on Computational Linguistics.
Cite (Informal):
Improving Accuracy of Low-resource ASR using Rule-Based Character Constituency Loss (RBCCL) (Ghimire et al., CHiPSAL 2025)
Copy Citation:
PDF:
https://aclanthology.org/2025.chipsal-1.6.pdf