Cross-Language Speech Synthesis using Transfer Learning

The advancement of Text-to-Speech (TTS) technology has opened new possibilities for personalized voice applications, particularly in multilingual contexts. This project focuses on developing a cross-language speech synthesis system, converting English text into high-quality Telugu audio. The primary challenge lies in retaining the unique vocal characteristics of a specific speaker during this language transformation. By leveraging advanced deep learning models, such as Tacotron 2 and WaveGlow, along with transfer learning techniques, the system addresses linguistic and phonetic differences to generate natural and intelligible Telugu speech. This work bridges the gap between diverse languages in speech synthesis while contributing to advancements in personalized and multilingual voice applications.