English

PolyNorm: Few-Shot LLM-Based Text Normalization for Text-to-Speech

Computation and Language 2025-11-06 v1 Machine Learning

Abstract

Text Normalization (TN) is a key preprocessing step in Text-to-Speech (TTS) systems, converting written forms into their canonical spoken equivalents. Traditional TN systems can exhibit high accuracy, but involve substantial engineering effort, are difficult to scale, and pose challenges to language coverage, particularly in low-resource settings. We propose PolyNorm, a prompt-based approach to TN using Large Language Models (LLMs), aiming to reduce the reliance on manually crafted rules and enable broader linguistic applicability with minimal human intervention. Additionally, we present a language-agnostic pipeline for automatic data curation and evaluation, designed to facilitate scalable experimentation across diverse languages. Experiments across eight languages show consistent reductions in the word error rate (WER) compared to a production-grade-based system. To support further research, we release PolyNorm-Benchmark, a multilingual data set covering a diverse range of text normalization phenomena.

Keywords

Cite

@article{arxiv.2511.03080,
  title  = {PolyNorm: Few-Shot LLM-Based Text Normalization for Text-to-Speech},
  author = {Michel Wong and Ali Alshehri and Sophia Kao and Haotian He},
  journal= {arXiv preprint arXiv:2511.03080},
  year   = {2025}
}

Comments

9 pages including appendix. EMNLP 2025 Industry Track

R2 v1 2026-07-01T07:22:11.287Z