A Novel Neural Machine Translation Approach for low-resource Sanskrit-Hindi Language pair

Sanskrit is one of the earliest native languages and correctly described as "the gods' language" because its wide use in Indian religious literature from past. However, it becoming less popular modern India. Due significant part to need for more materials translation both out Sanskrit, no longer commonly utilized. This study explores feasibility using machine (MT) provide a link between and, languages, contemporary descendant Hindi. A was conducted existing modelling methodologies, notably Statistical (SMT), proposed novel deep learning-based Machine strategy manually created parallel corpus Sanskrit-Hindi language pair. While SMT creates interpretations by mapping phrases source destination, statistical models, bilingual text corpora learning parameters, neural (NMT) frequently models entire single integrated model, convolutional network calculate probability word sequence. The NMT model implemented an encoder-decoder with attention mechanism paradigm inclusion gated recurrent units. Our approach involved development creation evaluated on automated human-based metrics, results show that our outperforms techniques Moses, surpassing them BLEU score 53.8% compared 34.56%. article examines undiscovered area Hindi discusses main benefits drawbacks while providing fresh viewpoint subject.