Translation of Different Dialects of Kannada to English Using Machine Learning

India is a Multilanguage country and Kannada is one of Indian historical language spoken by Millions of people. Kannada has a different versions from region to region. There exist different efficient algorithms to convert the formal language using a standard dictionary. Whereas converting the informal Kannada language to another language is a Challenging task, since each word of the language as a range of versions. The chief objective of the paper is to translate the informal Kannada language by understating the native vocabulary. This paper discusses about assessment approach to convert dialects of Kannada into English. For many native languages, particularly those with dialects, translation has been an unscathed field. When the source material contains dialects other than the official Kannada, existing translators fail to produce an accurate translation. This paper proposes LSTM algorithm to develop the translator. Automatic Speech Recognition (ASR) and other Translation models are used for accurate translation. The test pattern in this paper represents an improvement in the translation, with the validation accuracy of $86.32 \%$, thereby decreasing word Error Rate (WER).