ASR Error Correction in Low-Resource Burmese with Alignment-Enhanced Transformers using Phonetic Features

TL;DR

This study develops a Transformer-based error correction model for low-resource Burmese ASR, integrating phonetic and alignment features to significantly improve transcription accuracy.

Contribution

It introduces the first ASR error correction approach for Burmese, combining IPA and alignment features within Transformer models for enhanced performance.

Findings

WER reduced from 51.56 to 39.82 with AEC

chrF++ score improved from 0.5864 to 0.627

Consistent gains over baseline ASR outputs

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.