TaaS: Translation as a Service (Transformer Based Translator)

India with its rich linguistics is ranked as one of the top countries with diverse languages. According to the eighth schedule of the Constitution of India, there are about twenty-two scheduled languages, and various regional languages, making hundreds of languages spoken in various regions of the country. Communication is a key aspect of all parts of life, and it is time-consuming and difficult to master different languages. The linguistic nature of India is beautiful but at the same time poses a challenge to people for effective communication. As certain Indian languages have low language resources thus for this project low-resource environment is utilized for the design and development of the translator models. This would also help us to get an insight into the nature and efficiency of the trained models in specific domains, with low training and development resources. There are various accents of each language which differ from region to region. To minimize such complications the decision was taken to utilize the standard languages for the translation purpose. This paper focuses on developing translation models, under constrained resource using neural machine translation models, specifically transformers, on Indian Languages.