In a breakthrough that could transform communication for people with speech disabilities, researchers at the Indian Institute of Technology (IIT) Guwahati, in collaboration with Ohio State University, have developed an underwater vibration sensor that enables soundless, contactless voice recognition.
This innovation could pave the way for a new generation of assistive technologies, particularly for those who cannot use traditional voice-command systems due to vocal cord impairment or speech disorders.
The research, recently published in the journal Advanced Functional Materials, combines cutting-edge material science with artificial intelligence to detect and interpret the breath patterns produced when someone attempts to speak—even without generating sound.
At the heart of the innovation is a sponge-like, porous, and conductive material that reacts to chemical and mechanical changes. When a person exhales while attempting to speak, the airflow creates subtle ripples on the water’s surface. The sensor, placed just below this air-water interface, captures the micro-disturbances and converts them into electrical signals.
These signals are then processed using Convolutional Neural Networks (CNN), an AI-based method capable of recognizing complex patterns in the data. The result is a system that can “understand” silent speech with surprising accuracy.
“This is one of the rare material designs capable of enabling voice recognition purely through the detection of breath-induced water waves,” said Professor Uttam Manna, who led the study. “It opens up new possibilities for individuals who have partially or completely lost their ability to speak audibly.”
Voice recognition is a staple in modern digital interfaces, from smartphones and smart speakers to cars and home appliances. However, millions of people with speech impairments remain excluded from this convenience. Existing systems require audible speech and often fail to accommodate those with voice-related medical conditions such as laryngectomy, vocal cord paralysis, or severe stuttering.
The IIT Guwahati system fills this accessibility gap, offering a silent alternative that could potentially integrate with existing voice-assistive technologies and smart devices.
Currently, the lab-scale prototype costs around Rs 3,000, making it a relatively low-cost solution. The research team believes that with support from industry partners, the technology can be refined for commercial use and scaled for deployment in hospitals, rehabilitation centers, and even smart homes.
With proof-of-concept and AI integration complete, the researchers are now preparing for clinical trials to test the sensor’s efficacy in real-world medical scenarios. These trials will assess its usability by patients and healthcare professionals alike.