New Delhi, February 2, 2026: Recent research at the All India Institute of Medical Sciences (AIIMS), Delhi, suggests that speech patterns of an individual may provide objective clues for detecting depression at an early stage.
According to the World Health Organization (WHO), depression is distinct from the everyday mood changes and emotional fluctuations people commonly experience. It can affect multiple aspects of a person’s life, including relationships with family, friends, and the wider community.
The study is being conducted at an advanced Speech Health Lab established at AIIMS Delhi with corporate social responsibility (CSR) support. Researchers analyzed speech samples from 423 participants who had complete clinical and demographic records. The average age of participants was around 24 years, with the majority falling within the 18-25 age group.
Nearly two-thirds of participants were under 23, and about 75% were under 25. This indicates stronger engagement among younger individuals with low-barrier, speech-based mental health platforms. While the study included participants ranging from adolescence to older adulthood, participation steadily declined after the mid-30s.
Standard psychiatric screening identified that approximately 32% of participants had depressive symptoms. When these findings were compared with automated speech analysis, prediction accuracy ranged from 60% to 75%, increasing to nearly 78% when longer speech samples were assessed.
Researchers examined both linguistic and paralinguistic features of speech. These included fluency and articulation, as well as tone, pitch, emotional resonance, and vocal energy. Depression was found to reduce fluency, flatten prosody (the rhythm and melody of speech), and lower vocal effort.
“Analysis of speech offers a promising way to objectively identify signs of depression, as the cognitive and behavioral changes linked to the condition influence both the production and quality of speech,” said Dr. Nand Kumar, Professor, Department of Psychiatry, AIIMS Delhi. “Patients with depression often show reduced fluency, diminished prosody, or monotonous speech.”
Depression affects over 264 million people worldwide. India’s National Mental Health Survey 2015 found that one in 20 Indians experiences depressive disorders, with suicide identified as a major risk.
Early detection is particularly important among young people. A study by the National Institute of Mental Health and Neurosciences (NIMHANS) involving 8,542 college students across 15 cities found that one-third had moderate to severe depressive symptoms. The study also revealed that nearly one in five students reported suicidal thoughts.
Researchers emphasized that speech-based analysis is intended to support early screening and not replace clinical diagnosis. The model does not interpret emotions instead, it analyzes acoustic and quantifiable properties such as pitch variation, speech rate and pauses, intensity and energy levels, and prosody (intonation and rhythm).
All of these aspects can subtly change when a person is depressed, and machine learning models can identify patterns associated with depressive states.
Despite the promising results, researchers acknowledge important limitations. Accuracy can vary between individuals, languages, and recording contexts, and models require diverse training data to work effectively across different demographic groups. Ethical concerns also remain central to the development of such technologies. Issues related to privacy, consent, and the potential misuse of voice data must be carefully addressed before wider implementation.
(Rh/VK)
