BhaShammo: IPA Transcription of Bengali Regional Dialects using Dialect Guided Tokens

Authors: S M Jishanul Islam, Sadia Ahmmed, Sahid Hossain Mustakim, Mohammad Jawwadul Islam, Swakkhar Shatabda, Md. Rezuwan Hassan, Farig Sadeque

This is the official code repository for BhaShammo.

---

Getting Started

Running on Local Machines

Clone the repository

git clone https://github.com/S-M-J-I/dgt-regional-bengail-2-ipa.git

And install the packages

pip install requirements.txt

Running on Cloud

To run the notebooks on the cloud, download the training notebooks and the inference notebooks.

---

Model Checkpoints

The model checkpoints can be found through the HuggingFace collections.

---

Contributions

We do not accept any contributions for this repository unless it is from the Authors' side. Please raise an issue if you have any inquiries.

---

Acknowledgements

We would love to express our wholehearted gratitude towards the people in Bengali.AI for their support and supervision in this work and for providing the dataset. We would also love to thank them for arranging the national competition: ভাষামূল: মুখের ভাষার খোঁজে, Bengali Text with Regional Dialect to IPA (pronunciation) transcription, which gave birth to this idea and got acknowledged as the winning solution.

---