What Can Voice Analysis Tell You About Your Health?
Did you know that how you talk might actually say a lot about the state of your health and wellness?
A new hot topic in medical engineering research is disease and health detection via voice analysis.
There is growing research suggesting that voice analysis can help to assess health and diagnose certain diseases, and there is now a rush to develop technology that can do just that.
Voice analysis technology is of such high interest because it has the potential to be a reliable, efficient, affordable, convenient, and easy-to-use method to predict, diagnose, and support health concerns.1 Just think—what if your virtual home assistant (like Amazon’s Alexa) could let you know important information about your health in real time just by analyzing your voice?
How does voice analysis technology work?
Research is finding that not just what we say, but how we say it, might give us clues about our state of health.
It turns out that certain mental and physical health conditions are associated with changes in how you talk—such as variation in tone, rate of speech, slurring of words, emphases, length of pauses, and much more. Technologies are being developed that aim to predict and diagnose a wide variety of health conditions based on these kinds of factors.1 2 3
These technologies extract specific voice and acoustic features (referred to as vocal biomarkers) from audio voice samples using various techniques. Then, those features are analyzed for important patterns and cues to provide insights about your health.1 2
Over time, researchers hope to identify specific vocal biomarkers that can help diagnose, predict, and monitor certain health conditions.4
What health conditions could be linked to vocal biomarkers?
There is a growing body of evidence linking a wide range of mental and physical health conditions to certain vocal characteristics and features.
Here are just a few of the examples:
- Vocal tract disorders – Voice analysis can help detect abnormalities and pathologies in the vocal tract.5
- Mental health disorders like major depressive disorder, bipolar disorder, and schizophrenia – A study from UCLA looked at the voice patterns of people with these mental health disorders. The researchers were able to use those patterns to predict changes in mental health states over time quite accurately.3 One hope is that this kind of technology could help people get support sooner when they need it by picking up on negative changes immediately and connecting the patient with help.6
- Parkinson’s disease – Symptoms of Parkinson’s disease often include voice and speech problems. Voice analysis has the potential to accurately and easily test for Parkinson’s symptoms.7 8
- PTSD – One study found an artificial intelligence tool to be able to differentiate between the voices of people with and without PTSD with 89% accuracy.9 Factors like speaking rate, voice quality, and variation in voice collected through mobile sensing platforms have also been shown to predict symptoms of PTSD and depression.2
- Psychosis – Normal and dysfunctional thought patterns are associated with certain alterations in speech. Monitoring how things are said can support the diagnosis of thought disorders in psychosis.10
- Coronary artery disease – A Mayo Clinic study found a relationship between certain vocal characteristics and coronary artery disease.11
- Cardiac arrest – Researchers have found an audible biomarker that might be able to detect cardiac arrest. By analyzing audio gathered from remote, contactless devices, certain breathing patterns can be identified and emergency actions can be enacted to support the person in real time.12
Voice analysis for holistic health
Along with potentially being able to detect a wide variety of diseases and health conditions, voice analysis is already being used for preventative wellness and providing other individualized health & wellness information.
In fact, computerized analysis of acoustic vocal patterns to predict and manage states of health dates back more than 30 years. Some of the pioneers in this field include Dr. Sharry Edwards and Dr. Dorinne Davis.13 14
Expanding on the work of these earlier pioneers, ZYTO founder Dr. Vaughn Cook developed the EVOX system, which maps and analyzes a person’s voice when they speak about a specific topic. Through a patented biofeedback process, EVOX is able to provide to a person the information of the “missing” frequencies in their voice. This process helps the person shift their perception on the topic, which can lead to a variety of improvements in health, wellness, and performance.
How could voice analysis be used in the future?
There are countless possible uses and applications of health-related voice analysis technology. While ZYTO and other pioneers are already using this technology to assist with holistic health, there are now numerous companies, organizations, and startups getting in on the exciting scene of voice analysis—from IBM to the US Army.
One of the major goals of this area of research is to improve the diagnosis and detection of certain conditions. Many researchers, like Dr. Hosseini Ghomi who investigates vocal biomarkers of Parkinson’s disease, hope to create FDA-approved vocal biomarkers specific to certain disease processes. They want to create ways to reliably differentiate between different diseases (such as Parkinson’s, Alzheimer’s, ALS, etc.) to assist in better diagnosis.4
Voice analysis could help to detect certain diseases at earlier stages than current methods are able to. And it could also provide new biomarkers for diseases that are presently difficult to diagnose.1 7
The US Army is even getting involved, partnering with MIT researchers to work on detecting brain injuries using voice analysis.15
Another potential use of this technology would be to monitor patients remotely with apps or wearables to gather data on how symptoms change over time, how symptoms respond to medications, and more. Additionally, this kind of technology could help connect people with support services in real time when symptoms arise or health events occur.4
Ultimately, many developers of this technology hope to eventually move voice monitoring to the background, collecting data automatically and in real time using devices like smartphones, tablets, watches, or voice-controlled virtual assistants.2
The bottom line…
Voice analysis technology has the potential to differentiate between the voices of healthy and unhealthy people.1 This technology is already being successfully applied in a number of different medical and holistic health fields.
As the development of the technology progresses, voice analysis could introduce more efficiency, accuracy, convenience, and affordability to diagnosing, monitoring, and supporting various health conditions. The idea of having a sensor in your home or in your pocket at all times that could monitor your health and detect problematic health conditions by listening to your voice is quite intriguing.
1. Saloni, R.K. Sharma, & A.K. Gupta. “Disease detection using voice analysis: a review.” International Journal of Medical Engineering and Informatics 6, no. 3 (2014): 189-209.
2. Place, S., D. Blanch-Hartigan, C. Rubin, et al. “Behavioral Indicators on a Mobile Sensing Platform Predict Clinically Validated Psychiatric Symptoms of Mood and Anxiety Disorders.” Journal of Medical Internet Research 19, no. 3 (2017): e75.
3. Arevian, A.C., D. Bone, N. Malandrakis, et al. “Clinical state tracking in serious mental illness through computational analysis of speech.” PLOS ONE 15, no. 1 (2020): e0225695.
4. “Talk About a Revolution: The Future of Voice Biomarkers in the Neurology Clinic.” University of Washington. Depts.washington.edu.
5. Campisi, P., T.L Tewfik, J.J. Manoukian, et al. “Computer-Assisted Voice Analysis Establishing a Pediatric Database.” Arch Otolaryngology Head Neck Surg 128, no. 2 (2002): 156-160.
6. “AI in mental health screening: Voice analysis shows promise.” Healthline Media UK Ltd, Brighton, UK. Medicalnewstoday.com.
7. “Vision.” Parkinson’s Voice Initiative. Parkinsonsvoice.org.
8. Moro-Velázquez, L, J.A. Gómez-García, J.I. Godino-Llorente, et al. “Analysis of speaker recognition methodologies and the influence of kinetic changes to automatically detect Parkinson’s Disease.” Applied Soft Computing 62 (2018): 649-666.
9. Marmar, C.R., A.D. Brown, M. Qian, et al. “Speech‐based markers for posttraumatic stress disorder in US veterans.” Depression & Anxiety 36, no. 7 (2019): 607-616.
10. Mota, N.B., N.A.P. Vasconcelos, N. Lemos, et al. “Speech Graphs Provide a Quantitative Measure of Thought Disorder in Psychosis.” PLOS ONE 7, no. 4 (2012): e34928.
11. Maor, E., J.D. Sara, D. M. Orbelo, et al. “Voice Signal Characteristics Are Independently Associated With Coronary Artery Disease.” May Clinic Proceedings 93, no. 7 (2019): 840-847.
12. Chan, J., T. Rea, S. Gollakota, & J.E. Sunshine. “Contactless cardiac arrest detection using smart devices.” npj Digital Medicine 2 (2019): 52.
13. Edwards, Sharry. “Human bioacoustics biology: Acoustically anomalous vocal patterns used to detect biometric expressions relating to structural integrity and states of health.” The Journal of the Acoustical Society of America 120 (2006): 3286.
14. Davis, Dorinne S. The Cycle of Sound: A Missing Energetic Link. (New Pathways Press, 2012).
15. Crown, Ellen. “Army Partners with MIT Lincoln Lab on Voice Analysis Program to Detect Brain Injury.” U.S. Army. Army.mil.