Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Transcribing Bengali Text with Regional Dialects to IPA using District Guided Tokens

S M Jishanul Islam, Sadia Ahmmed, Sahid Hossain Mustakim

TL;DR

This work tackles Bengali text-to-IPA transcription across regional dialects by injecting district-specific information through District Guided Tokens (DGT). The authors fine-tune a byte-level transformer (ByT5) on a six-district Bengali dataset from the Bhashamul competition, demonstrating that DGT substantially improves performance over word-based models by mitigating OOV issues. ByT5 achieves the strongest results with public/private WERs around 2%, significantly better than baselines like umT5, BanglaT5, and mT5, highlighting the value of dialect-aware conditioning in phonetic transcription tasks. The findings suggest practical potential for dialect-aware NLP systems in Bengali and similar languages, with future work including ablations and models trained without DGT to assess its modular contribution.

Abstract

Accurate transcription of Bengali text to the International Phonetic Alphabet (IPA) is a challenging task due to the complex phonology of the language and context-dependent sound changes. This challenge is even more for regional Bengali dialects due to unavailability of standardized spelling conventions for these dialects, presence of local and foreign words popular in those regions and phonological diversity across different regions. This paper presents an approach to this sequence-to-sequence problem by introducing the District Guided Tokens (DGT) technique on a new dataset spanning six districts of Bangladesh. The key idea is to provide the model with explicit information about the regional dialect or "district" of the input text before generating the IPA transcription. This is achieved by prepending a district token to the input sequence, effectively guiding the model to understand the unique phonetic patterns associated with each district. The DGT technique is applied to fine-tune several transformer-based models, on this new dataset. Experimental results demonstrate the effectiveness of DGT, with the ByT5 model achieving superior performance over word-based models like mT5, BanglaT5, and umT5. This is attributed to ByT5's ability to handle a high percentage of out-of-vocabulary words in the test set. The proposed approach highlights the importance of incorporating regional dialect information into ubiquitous natural language processing systems for languages with diverse phonological variations. The following work was a result of the "Bhashamul" challenge, which is dedicated to solving the problem of Bengali text with regional dialects to IPA transcription https://www.kaggle.com/competitions/regipa/. The training and inference notebooks are available through the competition link.

Transcribing Bengali Text with Regional Dialects to IPA using District Guided Tokens

TL;DR

This work tackles Bengali text-to-IPA transcription across regional dialects by injecting district-specific information through District Guided Tokens (DGT). The authors fine-tune a byte-level transformer (ByT5) on a six-district Bengali dataset from the Bhashamul competition, demonstrating that DGT substantially improves performance over word-based models by mitigating OOV issues. ByT5 achieves the strongest results with public/private WERs around 2%, significantly better than baselines like umT5, BanglaT5, and mT5, highlighting the value of dialect-aware conditioning in phonetic transcription tasks. The findings suggest practical potential for dialect-aware NLP systems in Bengali and similar languages, with future work including ablations and models trained without DGT to assess its modular contribution.

Abstract

Accurate transcription of Bengali text to the International Phonetic Alphabet (IPA) is a challenging task due to the complex phonology of the language and context-dependent sound changes. This challenge is even more for regional Bengali dialects due to unavailability of standardized spelling conventions for these dialects, presence of local and foreign words popular in those regions and phonological diversity across different regions. This paper presents an approach to this sequence-to-sequence problem by introducing the District Guided Tokens (DGT) technique on a new dataset spanning six districts of Bangladesh. The key idea is to provide the model with explicit information about the regional dialect or "district" of the input text before generating the IPA transcription. This is achieved by prepending a district token to the input sequence, effectively guiding the model to understand the unique phonetic patterns associated with each district. The DGT technique is applied to fine-tune several transformer-based models, on this new dataset. Experimental results demonstrate the effectiveness of DGT, with the ByT5 model achieving superior performance over word-based models like mT5, BanglaT5, and umT5. This is attributed to ByT5's ability to handle a high percentage of out-of-vocabulary words in the test set. The proposed approach highlights the importance of incorporating regional dialect information into ubiquitous natural language processing systems for languages with diverse phonological variations. The following work was a result of the "Bhashamul" challenge, which is dedicated to solving the problem of Bengali text with regional dialects to IPA transcription https://www.kaggle.com/competitions/regipa/. The training and inference notebooks are available through the competition link.
Paper Structure (10 sections, 1 equation, 2 figures, 2 tables)

This paper contains 10 sections, 1 equation, 2 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Dataset Analysis
  3. Methodology
  4. The District Guided Tokens
  5. Transformer Model
  6. Experiments
  7. Experimental Setup
  8. Evaluation Metrics
  9. Results and Analysis
  10. Conclusion

Figures (2)

  • Figure 1: The distribution of sentences in the train and test sets.
  • Figure 2: An overview of the District Guided Token methodology.