Guwahati: Researchers at IIT-Guwahati have developed an underwater vibration sensor that enables automated and contactless voice recognition.

Developed in collaboration with Ohio State University in the US, the sensor offers a promising alternative communication method for individuals with voice disabilities who are unable to use conventional voice-based systems, IIT-Guwahati said in a statement on Monday.

Voice recognition has become an integral part of modern life, helping users operate smart devices, including mobile phones and home appliances, through voice commands. However, for people with voice disorders, these technologies remain inaccessible.

To address this limitation, the research team focused on the air exhaled through the mouth while speaking. In cases where individuals cannot produce sound, attempting to speak generates airflow from the lungs. When this air flows over a water surface, it produces subtle waves.

The team developed an underwater vibration sensor which can detect these water waves and interpret speech signals without depending on audible voice, creating a new pathway for voice recognition. Made from a conductive, chemically reactive porous sponge, the sensor is placed just below the air-water interface and captures the tiny disturbances created by exhaled air, converting them into measurable electrical signals.

Using Convolutional Neural Networks (CNN), a type of deep learning model, the system accurately recognises subtle signal patterns, allowing users to communicate with devices from a distance without needing to produce sound. The findings have been published in the journal Advanced Functional Materials.

“It is one of the rare material designs that recognises voice by monitoring water waves formed at the air-water interface due to exhaled air from the mouth. This approach is likely to provide a viable solution for communication with individuals with partially or entirely damaged vocal cords,” said Professor Uttam Manna, a member of the research team.

The working prototype currently costs around Rs 3,000 at lab scale. With industry collaboration, the final product is expected to be more affordable.

Key features of the developed sensor include AI-powered interpretation using CNNs and hands-free control of smart devices. The researchers are now planning to seek clinical validation for the device.