Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2511.03080 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914137387302912 |
|---|---|
| author | Wong, Michel Alshehri, Ali Kao, Sophia He, Haotian |
| author_facet | Wong, Michel Alshehri, Ali Kao, Sophia He, Haotian |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_03080 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | PolyNorm: Few-Shot LLM-Based Text Normalization for Text-to-Speech Wong, Michel Alshehri, Ali Kao, Sophia He, Haotian Computation and Language Machine Learning 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. |
| title | PolyNorm: Few-Shot LLM-Based Text Normalization for Text-to-Speech |
| topic | Computation and Language Machine Learning |
| url | https://arxiv.org/abs/2511.03080 |