Abstract: Parallel dataset which needs large number of aligned sources and target language pairings have traditionally been used to develop speech to speech translation system. This study presents a ...