BanglaIPA: Towards Robust Text-to-IPA Transcription with Contextual Rewriting in Bengali

TL;DR

BanglaIPA tackles the challenge of robust Bengali text-to-IPA transcription across standard language, six regional dialects, and numerals by integrating contextual numeral rewriting with a Transformer-based IPA generator and a State Alignment mechanism that handles out-of-vocabulary segments. The system introduces a Word-IPA dictionary that caches transcriptions and reduces repeated computation, and it uses a context-aware rewriting stage powered by a large language model to ensure numerals are pronounced appropriately in context. Empirical results on the DUAL-IPA benchmark show an overall mean word error rate of $11.4\%$, with improvements of between $58.4\%$ and $78.7\%$ over strong baselines, and strong cross-dialect performance (e.g., Rangpur at $10.4\%$). The work advances Bengali TTS/ASR pipelines by providing an end-to-end, regionally robust IPA transcription framework and introduces the STAT algorithm for robust subword handling, while outlining limitations such as dialect coverage and the computational cost of LLM-based numeral rewriting.

Abstract

Despite its widespread use, Bengali lacks a robust automated International Phonetic Alphabet (IPA) transcription system that effectively supports both standard language and regional dialectal texts. Existing approaches struggle to handle regional variations, numerical expressions, and generalize poorly to previously unseen words. To address these limitations, we propose BanglaIPA, a novel IPA generation system that integrates a character-based vocabulary with word-level alignment. The proposed system accurately handles Bengali numerals and demonstrates strong performance across regional dialects. BanglaIPA improves inference efficiency by leveraging a precomputed word-to-IPA mapping dictionary for previously observed words. The system is evaluated on the standard Bengali and six regional variations of the DUAL-IPA dataset. Experimental results show that BanglaIPA outperforms baseline IPA transcription models by 58.4-78.7% and achieves an overall mean word error rate of 11.4%, highlighting its robustness in phonetic transcription generation for the Bengali language.

Figures (6)

• Figure 1: Existing IPA transcription systems (left) are limited to standard Bengali and struggle to process texts containing regional dialectal variations and numerical expressions. In contrast, BanglaIPA (right) successfully transcribes both standard and regional dialects and accurately handles numerical expressions through the incorporation of a contextual rewriting procedure.
• Figure 2: Overview of the end-to-end processing pipeline of the BanglaIPA system for generating the International Phonetic Alphabet (IPA) transcription.
• Figure 3: Region-wise word error rate distribution on the DUAL-IPA test dataset, with IPA transcriptions generated by the BanglaIPA system. The system consistently shows very low word error rates across most regions.
• Figure 4: The pronunciation of Bengali numbers varies depending on the textual context (highlighted in colors). The BanglaIPA system converts numbers into word forms as an intermediate step in IPA transcription.
• Figure 5: Identification of the optimal Elbow Point (EP) to terminate the dataset splitting procedure for the Rangpur dialect using the Elbow method.