A SHORT voice note may soon help identify depression, as new research shows machine learning can detect depressive profiles from everyday speech with high accuracy. The findings suggest machine learning could support earlier and more accessible mental health screening.
Voice Patterns and Depression Screening Challenges
Depression affects more than 280 million people worldwide and remains widely underdiagnosed due to the absence of reliable, objective biomarkers. Subtle changes in speech and acoustic features are known to reflect emotional and cognitive states, making voice a promising source of diagnostic signals. Machine learning offers a way to analyse these complex patterns at scale, potentially distinguishing between adaptive and non-adaptive mood profiles using natural, real-world communication such as WhatsApp audio messages.
Machine Learning Models Tested on WhatsApp Audio
Researchers evaluated seven machine learning models using WhatsApp audio recordings from 160 Brazilian Portuguese speakers, including individuals with major depressive disorder and healthy controls. Participants were divided into separate training and testing cohorts. Audio samples included both structured tasks such as counting from one to ten and semi structured speech describing the past week. After preprocessing, 68 acoustic features were extracted and used to train the models. In the test group, diagnoses were confirmed using the Mini International Neuropsychiatric Interview.
Machine learning models achieved peak accuracies of 91.67% for women and 80% for men, with an area under the curve of 91.9% for women and 78.33% for men. Performance varied depending on the audio instruction type, with higher accuracy seen in spontaneous speech. When using structured counting tasks, accuracy reached 82% for women and 78% for men. These results demonstrate that machine learning can classify whether a WhatsApp audio message represents a depressive patient or a healthy individual with reasonable reliability.
Clinical Implications and Future Directions
The findings highlight the potential of machine learning as a low cost, low burden screening tool that aligns with everyday communication habits. Such approaches could support clinicians by identifying individuals who may benefit from further assessment, particularly in settings with limited mental health resources.
Reference
Otani VHO et al. ML-based detection of depressive profile through voice analysis in WhatsApp audio messages of Brazilian Portuguese Speakers, PLOS Mental Health. 2026;DOI:10.1371/journal.pmen.0000357.
