How Pranet Khetan built India’s first Hindi speech recognition tool for patients
Pranet Khetan is a 16-year-old inventor, who has created Paraspeak, an AI assisted device that helps patients who have neurological problems, decode slurred Hindi into clear audio. The gadget is inexpensive, wearable, and constructed on the foundation of transformer-based speech recognition.
This video is not available.
When most high school students are concerned with examinations and college applications, 16 year old Pranet Khetan is inventing technology that will change the lives of thousands. It is an AI, self-learning, miniature device invented by a Class 11 student of Shiv Nadar School, Gurgaon, to make people with speech disorders as a result of paralysis and neurological disorders communicate effectively.
During our interview, Pranet told us that he had the idea when on a school field trip in a paralysis care centre and when he realized how hard it was for the patients to speak. He needed to do something about it, so he spent months developing the solution that supported it with artificial intelligence and speech recognition and, what was important, the solution was tailored to the Hindi speakers.
Recognised Globally
His research has already won him at the IRIS National Fair in India and Regeneron International Science and Engineering Fair (ISEF) 2025 in the United States. To him, the actual reward is enabling people to speak once more. The reason is that, according to him, it is the ability to create a physical human impact and change lives of people using my skills that motivates him.
At present, Paraspeak is among the existing few AI-based solutions catering towards dysarthric patients speaking Hindi. With the further development, it may be able to support more languages and transform the assistive care in India and other regions of the world.
A solution that’s scalable and affordable
The most common use of speech tools is customized according to a particular patient. Nevertheless, ParaSpeak is not only designed to work with a lot of users of different types of diseases. This makes it scalable especially in a country like India where the needs are different and healthcare technology is not readily available.
All the device is made out of custom PCBs and the entire device can be manufactured at only Rs 2000 plus a small monthly internet bill. Pranet indicated that it was interesting to learn the method of PCB design. I employed my knowledge in technology, took very long to figure out EDA software, and became familiar with all the tools at my disposal, and by trial and error, and YouTube videos, were very helpful.”