Researchers at the Sardar Vallabhbhai National Institute of Technology in Surat and Germany’s Rhenish University of Applied Science developed a method to detect speech signals in a non-invasive manner.
Most of us have may have come across instances when we suspect that certain colleagues are faking an illness to call in sick at work. Now, researchers have developed an AI (artificial intelligence) tool to detect whether someone has a cold by the tone of their voice.
Researchers at the Sardar Vallabhbhai National Institute of Technology in Surat and Germany’s Rhenish University of Applied Science developed a method to detect speech signals in a non-invasive manner.
They analysed the voice samples of 630 German people,111 of whom had a cold, according to a Business Insider report based on a feature by The Economist. They captured variations in the voice spectrum and reportedly studied the vocal rhythms found in human speech.
The analysis showed that they could effectively classify cold and non-cold speech.
“The proposed features efficiently capture the spectral difference between cold and non-cold speech and can be utilised to automatically diagnose common cold and related disorders,” the researches said in a study published in science journal ScienceDirect.
The aim of the study was to develop a method that gives higher performance with fewer features to identify a person with a common cold from their speech.
They were asked to count from one to 40 and to describe what they did on the weekend, The Economist reported. There were also asked to recite Aesop’s fable “The North Wind and the Sun”.
The researches feel that the feature can help people prevent the spread of viral infections and remotely monitor the health of a person with a common cold.