ASR Error Correction in Low-Resource Burmese with Alignment-Enhanced Transformers using Phonetic Features
Ye Bhone Lin, Thura Aung, Ye Kyaw Thu, Thazin Myint Oo
TL;DR
This work tackles automatic speech recognition error correction for low-resource Burmese by proposing an alignment-guided Transformer that incorporates IPA-based phonetic features and alignment information. Using two open Burmese corpora and a data augmentation pipeline, the authors show that AEC consistently improves WER and chrF++ across multiple ASR backbones, with alignment features delivering especially strong gains. The study also analyzes augmentation effects, revealing distribution-shift challenges for strong models but potential benefits for smaller models in character-level similarity. The findings highlight the robustness of AEC in resource-scarce settings and suggest practical directions for feature design and future integration with language models to further enhance post-ASR correction.
Abstract
This paper investigates sequence-to-sequence Transformer models for automatic speech recognition (ASR) error correction in low-resource Burmese, focusing on different feature integration strategies including IPA and alignment information. To our knowledge, this is the first study addressing ASR error correction specifically for Burmese. We evaluate five ASR backbones and show that our ASR Error Correction (AEC) approaches consistently improve word- and character-level accuracy over baseline outputs. The proposed AEC model, combining IPA and alignment features, reduced the average WER of ASR models from 51.56 to 39.82 before augmentation (and 51.56 to 43.59 after augmentation) and improving chrF++ scores from 0.5864 to 0.627, demonstrating consistent gains over the baseline ASR outputs without AEC. Our results highlight the robustness of AEC and the importance of feature design for improving ASR outputs in low-resource settings.