LLM-Powered Grapheme-to-Phoneme Conversion: Benchmark and Case Study

TL;DR

This paper evaluates large language models for grapheme-to-phoneme conversion, demonstrating their potential to outperform traditional tools through prompting and post-processing, especially in underrepresented languages like Persian.

Contribution

It introduces a benchmarking dataset and methods to enhance LLM-based G2P conversion without additional training, highlighting their effectiveness in complex phonetic scenarios.

Findings

LLMs can outperform traditional G2P tools with proper prompting.

Proposed methods improve G2P accuracy without extra training.

Effective in underrepresented languages like Persian.

Abstract

Grapheme-to-phoneme (G2P) conversion is critical in speech processing, particularly for applications like speech synthesis. G2P systems must possess linguistic understanding and contextual awareness of languages with polyphone words and context-dependent phonemes. Large language models (LLMs) have recently demonstrated significant potential in various language tasks, suggesting that their phonetic knowledge could be leveraged for G2P. In this paper, we evaluate the performance of LLMs in G2P conversion and introduce prompting and post-processing methods that enhance LLM outputs without additional training or labeled data. We also present a benchmarking dataset designed to assess G2P performance on sentence-level phonetic challenges of the Persian language. Our results show that by applying the proposed methods, LLMs can outperform traditional G2P tools, even in an underrepresented language like Persian, highlighting the potential of developing LLM-aided G2P systems.