Guwahati: Researchers from the Sardar Vallabhbhai National Institute of Technology in Surat, India, and Germany’s Rhenish University of Applied Science have developed an AI tool to detect whether someone has a cold just from the sound of their voice.
This non-invasive method analyses voice patterns using harmonics to determine if a person is truly suffering from a cold.
The study analysed the voice patterns of 630 people, of which 111 had a cold, and tested their speech patterns to classify cold and non-cold speech.
Also Read: Bengaluru firm to fund employees’ signing for ChatGPT Plus subscriptions
Volunteers were asked to count from 1 to 40, describe their weekend activities, and recite Aesop’s fable, The North Wind and the Sun. The study’s accuracy in detecting cold was 70%.
The proposed features in the study capture the spectral difference between cold and non-cold speech efficiently, enabling automatic diagnosis of the common cold and related disorders.
This method could be used to prevent the spread of viral infections and remotely monitor the health of people with a common cold.
Also Read: OpenAI unveils customizable Chatbot feature to address AI bias concerns
The development of this AI tool aims to identify people with a common cold using fewer features, thus giving higher performance.
This research may lead to future developments in detecting other illnesses and diseases by analysing voice patterns.