About
Anusha Prakash recently graduated as a PhD Scholar from the Electrical Engineering department at the Indian Institute of Technology (IIT) Madras, India. Her research focused on developing high-quality text-to-speech (TTS) synthesizers for Indian languages, with a particular emphasis on low-resource languages in a multilingual context. She has extensive experience working with various TTS frameworks, ranging from traditional techniques like unit selection synthesis, HMM-based methods, and conventional neural networks to state-of-the-art end-to-end speech synthesizers (including transformer-based FastSpeech(2) networks). Her work integrates linguistic knowledge and signal processing techniques with deep learning methods to enhance synthesis quality and intelligibility. Additionally, she has contributed to improving the quality of dysarthric speech.
Anusha has been involved in various government-funded projects. She served as a Principal Project Officer in a project titled “Speech Technologies for Indian Languages,” which is part of the “National Language Translation Mission (NLTM)-Bhashini.” The primary goal of this project is to translate technical lectures in English into various Indian languages. She has trained TTS systems for 15 Indian languages, including Assamese, Bengali, Bodo, Gujarati, Hindi, Kannada, Malayalam, Manipuri, Marathi, Odia, Punjabi, Rajasthani, Tamil, Telugu, and Urdu, as well as Indian English with different accents. Some of these models have been deployed as part of the project and have been integrated with other applications, such as speech-to-speech and video-to-video transcreations.