Voice Analysis Software Shows Promise as PTSD Diagnosis Tool for Veterans
May 14, 2019
Tampa, FL (Law Firm Newswire) May 14, 2019 – Researchers have developed voice analysis software that can help identify post-traumatic stress disorder (PTSD) in veterans by analyzing their speech. A recent study used an artificial intelligence (AI) program to analyze vocal characteristics such as rhythm and tone after training it on how to classify people based on speech examples.
According to the findings, the AI could distinguish which individuals had PTSD and which did not with 89 percent accuracy. Researchers recorded hours-long speech samples from 53 Afghanistan and Iraq veterans with PTSD linked to their service. They also interviewed 78 former service members who did not have the mental health condition.
The samples were processed by voice software from Stanford Research Institute (SRI) International, which is the company that developed Apple’s Siri voice assistant. The software produced 40,526 speech segments for analysis.
“The research seems promising for easier, more accurate PTSD diagnosis in the future,” commented Florida veterans lawyer David W. Magann. “Such technology can help facilitate early intervention to identify veterans who need to be evaluated for PTSD. Veterans who are diagnosed with it can then seek helpful treatments and receive the benefits they need rather than struggling with their mental health.”
The software linked specific speech patterns with PTSD. The voices of veterans with PTSD were identified as having unclear speech, longer hesitations and a “lifeless, metallic tone.” Such voice features have long been thought to help diagnose the mental condition.
According to Dimitra Vergyri, head of the Speech Technology and Research Laboratory at SRI International, the AI uses speech analysis to determine an individual’s state in terms of how they are feeling, their thought patterns and quality of communication, as well as their physical and mental health.
The study did not examine the science behind PTSD and vocal patterns. However, researchers noted the existence of theories on how a person’s speech can be affected when traumatic events alter the way the brain processes emotions. They are aiming to continue training and enhancing the AI tool in order to qualify for government permission to use the software clinically in the future.
