Non-Standard Vietnamese Word Detection and Normalization for Text-to-Speech

TL;DR

This paper presents a two-phase approach for Vietnamese non-standard word detection and normalization in TTS systems, utilizing advanced models and rule-based algorithms to improve accuracy and handle diverse NSW types.

Contribution

It introduces a novel combination of model-based tagging and rule-based normalization specifically tailored for Vietnamese NSWs in TTS applications.

Findings

BiLSTM-CNN-CRF and BERT-BiGRU-CRF models achieve over 90% F1 scores.

The approach reduces sentence error rates to below 8%.

BERT-BiGRU-CRF yields the highest F1 score of 95%.

Abstract

Converting written texts into their spoken forms is an essential problem in any text-to-speech (TTS) systems. However, building an effective text normalization solution for a real-world TTS system face two main challenges: (1) the semantic ambiguity of non-standard words (NSWs), e.g., numbers, dates, ranges, scores, abbreviations, and (2) transforming NSWs into pronounceable syllables, such as URL, email address, hashtag, and contact name. In this paper, we propose a new two-phase normalization approach to deal with these challenges. First, a model-based tagger is designed to detect NSWs. Then, depending on NSW types, a rule-based normalizer expands those NSWs into their final verbal forms. We conducted three empirical experiments for NSW detection using Conditional Random Fields (CRFs), BiLSTM-CNN-CRF, and BERT-BiGRU-CRF models on a manually annotated dataset including 5819 sentences extracted from Vietnamese news articles. In the second phase, we propose a forward lexicon-based maximum matching algorithm to split down the hashtag, email, URL, and contact name. The experimental results of the tagging phase show that the average F1 scores of the BiLSTM-CNN-CRF and CRF models are above 90.00%, reaching the highest F1 of 95.00% with the BERT-BiGRU-CRF model. Overall, our approach has low sentence error rates, at 8.15% with CRF and 7.11% with BiLSTM-CNN-CRF taggers, and only 6.67% with BERT-BiGRU-CRF tagger.