# g2pW: A Conditional Weighted Softmax BERT for Polyphone Disambiguation in Mandarin

arXiv: https://arxiv.org/abs/2203.10430 · HTML: https://mlsanity.com/paper/2203.10430/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Builds on 4 of its 18 references that are in the map.
- First posted on arXiv 20 Mar 2022.

## Builds on (most cited first, 4 of 4)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [g2pM: A Neural Grapheme-to-Phoneme Conversion Package for Mandarin Chinese Based on a New Open Benchmark Dataset](https://mlsanity.com/paper/2004.03136/) (2020)
- [Polyphone Disambiguation for Mandarin Chinese Using Conditional Neural Network with Multi-level Embedding Features](https://mlsanity.com/paper/1907.01749/) (2019)
- [Polyphone Disambiguation in Mandarin Chinese with Semi-Supervised Learning](https://mlsanity.com/paper/2102.00621/) (2021)

## Built on by (most cited first, 7 of 7)

- [The taste of IPA: Towards open-vocabulary keyword spotting and forced alignment in any language](https://mlsanity.com/paper/2311.08323/) (2023)
- [Let's Fuse Step by Step: A Generative Fusion Decoding Algorithm with LLMs for Robust and Instruction-Aware ASR and OCR](https://mlsanity.com/paper/2405.14259/) (2024)
- [Multilingual context-based pronunciation learning for Text-to-Speech](https://mlsanity.com/paper/2307.16709/) (2023)
- [External Knowledge Augmented Polyphone Disambiguation Using Large Language Model](https://mlsanity.com/paper/2312.11920/) (2023)
- [Dict-TTS: Learning to Pronounce with Prior Dictionary Knowledge for Text-to-Speech](https://mlsanity.com/paper/2206.02147/) (2022)
- [VocalRender: Score-Native Singing Voice Synthesis for Real-World Composition](https://mlsanity.com/paper/2607.27768/) (2026)
- [Dictionary-Constrained Grapheme-to-Phoneme for Unsegmented Languages from LLM-Annotated Data](https://mlsanity.com/paper/2609.19805/) (2026)

## Abstract (arXiv)

Polyphone disambiguation is the most crucial task in Mandarin grapheme-to-phoneme (g2p) conversion. Previous studies have approached this problem using pre-trained language models, restricted output, and extra information from Part-Of-Speech (POS) tagging. Inspired by these strategies, we propose a novel approach, called g2pW, which adapts learnable softmax-weights to condition the outputs of BERT with the polyphonic character of interest and its POS tagging. Rather than using the hard mask as in previous works, our experiments show that learning a soft-weighting function for the candidate phonemes benefits performance. In addition, our proposed g2pW does not require extra pre-trained POS tagging models while using POS tags as auxiliary features since we train the POS tagging model simultaneously with the unified encoder. Experimental results show that our g2pW outperforms existing methods on the public CPP dataset. All codes, model weights, and a user-friendly package are publicly available.