Back-Translation-Style Data Augmentation for Mandarin Chinese Polyphone Disambiguation

TL;DR

This paper introduces a back-translation-style data augmentation approach for Mandarin Chinese polyphone disambiguation, leveraging unlabeled text data and dual G2P and P2G models to improve accuracy in TTS systems.

Contribution

It proposes a novel data augmentation method using back-translation techniques to enhance polyphone disambiguation models with unlabeled data.

Findings

Improved polyphone disambiguation accuracy on benchmark datasets.

Effective handling of data scarcity and imbalanced polyphone distributions.

Demonstrated the method's superiority over traditional training approaches.

Abstract

Conversion of Chinese Grapheme-to-Phoneme (G2P) plays an important role in Mandarin Chinese Text-To-Speech (TTS) systems, where one of the biggest challenges is the task of polyphone disambiguation. Most of the previous polyphone disambiguation models are trained on manually annotated datasets, and publicly available datasets for polyphone disambiguation are scarce. In this paper we propose a simple back-translation-style data augmentation method for mandarin Chinese polyphone disambiguation, utilizing a large amount of unlabeled text data. Inspired by the back-translation technique proposed in the field of machine translation, we build a Grapheme-to-Phoneme (G2P) model to predict the pronunciation of polyphonic character, and a Phoneme-to-Grapheme (P2G) model to predict pronunciation into text. Meanwhile, a window-based matching strategy and a multi-model scoring strategy are proposed to judge the correctness of the pseudo-label. We design a data balance strategy to improve the accuracy of some typical polyphonic characters in the training set with imbalanced distribution or data scarcity. The experimental result shows the effectiveness of the proposed back-translation-style data augmentation method.