Efficient Dialect-Aware Modeling and Conditioning for Low-Resource Taiwanese Hakka Speech Processing

TL;DR

This paper introduces a dialect-aware RNN-T framework for Taiwanese Hakka speech recognition, effectively disentangling dialectal variation from linguistic content to improve accuracy across scripts and dialects.

Contribution

It presents the first unified model that jointly handles dialectal variability and dual writing systems in low-resource Hakka ASR, leveraging parameter-efficient strategies.

Findings

Achieved 57.00% relative error reduction on Hanzi ASR

Achieved 40.41% relative error reduction on Pinyin ASR

First systematic study of Hakka dialectal effects on ASR

Abstract

Taiwanese Hakka is a low-resource, endangered language that poses significant challenges for automatic speech recognition (ASR), including high dialectal variability and the presence of two distinct writing systems (Hanzi and Pinyin). Traditional ASR models often encounter difficulties in this context, as they tend to conflate essential linguistic content with dialect-specific variations across both phonological and lexical dimensions. To address these challenges, we propose a unified framework grounded in the Recurrent Neural Network Transducers (RNN-T). Central to our approach is the introduction of dialect-aware modeling strategies designed to disentangle dialectal "style" from linguistic "content", which enhances the model's capacity to learn robust and generalized representations. Additionally, the framework employs parameter-efficient prediction networks to concurrently model ASR (Hanzi and Pinyin). We demonstrate that these tasks create a powerful synergy, wherein the cross-script objective serves as a mutual regularizer to improve the primary ASR tasks. Experiments conducted on the HAT corpus reveal that our model achieves 57.00% and 40.41% relative error rate reduction on Hanzi and Pinyin ASR, respectively. To our knowledge, this is the first systematic investigation into the impact of Hakka dialectal variations on ASR and the first single model capable of jointly addressing these tasks.