IRJET - Speech to Speech Translation using Encoder Decoder Architecture

This document summarizes a research paper on speech-to-speech translation using an encoder-decoder architecture. It describes a system that takes speech in one language as input, recognizes the speech to generate text, translates the text to another language, and synthesizes speech in the other language as output. The system consists of three main modules: speech recognition in the source language, text translation between languages, and speech generation in the target language. It aims to enable two-way translation between spoken sentences in different languages.