Popular GitHub Repositories for Bengali

BNLP is a natural language processing toolkit for Bengali Language.

This repository contains the official release of the model "BanglaBERT" and associated downstream finetuning code and datasets introduced in the paper titled "BanglaBERT: Language Model Pretraining and Benchmarks for Low-Resource Language Understanding Evaluation in Bangla" accpeted in Findings of the Annual Conference of the North American Chapter of the Association for Computational Linguistics: NAACL-2022.

Bangla-Bert is a pretrained bert model for Bengali language

Towards Building Text-To-Speech Systems for the Next Billion Users - Microsoft Research Intern Work - Accepted at ICASSP 2023

Fully Configurable RAG Pipeline for Bengali Language RAG Applications. Supports both Local and Huggingface Models, Built with Langchain.

Bangla NLP dataset. Bangla NER,POStag, text summarization, stopword, translate, sentiment analysis, wiki articles, root word, dataset etc.

Aiming to achieve ultimate Multilingual TTS pipeline with main focus on releasing COQUI?TTS(Text-to-Speech) based high performing neural voice cloning systems for Bangla for the first time, supporting different SOTA models for Bangla and also Multilingual (Arabic+Bengali) code mixed TTS pipeline.

An easy-to-use Python module that helps you to extract the BERT embeddings for a large text dataset (Bengali/English) efficiently.

Machine Translation using Transfromers

MediBeng Whisper Tiny 通过训练 Whisper Tiny 模型来翻译孟加拉语和英语混合语音，将其转换成英语，从而改进医生与患者对话的转录工作。这使得分析、记录保存以及在医疗保健中使用人工智能变得更加容易。

BanglaTTS is a text-to-speech (TTS) system for Bangla language that works in offline mode. You can convert text to speech in male or female voice.

Classification Benchmarks for Under-resourced Bengali Language based on Multichannel Convolutional-LSTM Network

Fine tuned llama 3 models for context based question answering in bengali language.

Train and finutune text-to-speech models for Bengali and many other languages!

Bangla News Article Categorization Using Conv-LSTM Net. It is a multi-class classification problem.

Goal of this work is to take Bengali one or more words as input in a system and predict the next most likely word and also suggest the full possible sentence as output. Recurrent Neural Network (RNN) was used with Gated Recurrent Unit (GRU) to train and create the model. About 200,000 plus word data has been used as dataset. The dataset has been collected from several authentic sources. We also published a paper based on our unique work.

A Bengali Text Classification Task Using Deep Learning. It is Multi-class Text Classification Problem. Bengali news headlines used for the categorization.

The work presents “Anwesha: A tool for Semantic Search in Bangla”. This work shows explorations toward building a search engine prototype in Bangla language. The work used sources from WordNet, Wikipedia and statistical co-occurences(LSA) for retrieval of semantically related documents.

Bengali News Summarization - BengaliGPT & T5

Bengali word embedding using BengaliWord2Vec from BNLP. A mini project under the mentorship of Prof. Sandipan Ganguly, HIT-K.