The SYSPIN project from SPIRE Lab, IISc, Bangalore, in collaboration with Bhashini AI Solutions, has released TTS corpuses in 9 Indian languages with 2 speakers (male and female) per language. Over 40 hours of speech per speaker were recorded, ensuring high quality through rigorous checks, including local and native speaker contributions. The SYSPIN dataset and baseline TTS models are now available for download, empowering innovations in sectors like agriculture, healthcare, education, and finance. SYSPIN aims to advance multilingual, multi-speaker TTS systems, and organized challenges under LIMMITS 23, 24, and 25 to support inclusive, accessible voice technologies in India.

Learn more about the project

Speech recognition in agriculture and finance for the poor is an initiative predominantly to create resources and make them available as a digital public good in the open source domain to spur research and innovation in speech recognition in nine different Indian languages in the area of agriculture and finance. Nine Indian languages considered for this project are Hindi, Bengali, Marathi, Telugu, Bhojpuri, Kannada, Magadhi, Chhattisgarhi, and Maithili.

Learn more about the project

This project aims to address the gap between technological advancements and clinical intervention tools by developing automatic/semi-automatic analytical tools for speech and voice data that can be used to assist in the objective, quantitatively based diagnosis and treatment of speech and language difficulties associated with Autism Spectrum Disorder (ASD) and make it accessible to individuals of a variety of linguistic, cultural, and socioeconomic backgrounds in India.

'English Gyani' aims at improving one's English comprehension and reading skill with guidance in vernacular to cater to the increasing demand of English learning particularly for employment and promotions of individuals under-resourced to avail English learning materials or schools. The lessons are categorized to improve the vocabulary, grammar and sentence construction skills as parts of English comprehension.

Learn more about the project

One of the typical symptoms of neuro-degenerative diseases like Amyotrophic Lateral Sclerosis (ALS), Parkinson's Disease (PD) and Alzheimer's Disease is dysarthria, that is, difficulty in speech production. This project aims to develop robust algorithms for early detection and monitoring of neuro-degenerative diseases using speech cues and for enhancement of the naturalness and intelligibility of the dysarthric speech.

Learn more about the project

This project aims to develop a robust keyword spotting system for VoIP speech in real time which involves several unique challenges that demand very fast processing of concurrent sessions, high accuracy and minimal false alarms in the outputs, handling an unrestricted vocabulary and robust performance to codec and channel variations.

Learn more about the project

Understanding of how various linguistic and paralinguistic components are exchanged in a human speech communication system could potentially be useful for making human-machine interaction natural.

Developing models for synthesizing facial expressions in various affective states would be useful for creating a natural agent in a human-machine communication, which could be used in a variety of applications including customer support which requires presence of an agent for interacting with the customers.

Exploring the brain mechanisms associated with micro-sleep architecture dynamics among healthy practitioners of Vipassana meditation (who differ in their meditation proficiency) and control subjects.

Developing a supervised system that predicts quantitative scores corresponding to the degree of nativity in the Indian spoken English and provides qualitative feedback according to the learners' nativity in an automated way.