If a friend or relative calls you on the phone, and something is wrong in his or her life, you can tell immediately. Could an AI tell also? Yes, and a whole lot more. Using the same machine learning technology an AI can use to tell a cat from a dog or recognize a person’s face, an AI could recognize your voice. An AI with access to a database containing a large number of your voice samples along with a description of whose voice each sample is associated with, the AI can recognize you. Based on characteristics of the voice and what state of mind those characteristics are associated with, it can also know if you are not feeling well, are upset about something, or in a big hurry. Some of the characteristics of voice data which can be detected include tone, tempo, volume, language and dialect used, and other voice characteristics.
John McCormick, deputy editor of WSJ Pro Artificial Intelligence, wrote an excellent article about voice recognition called, “What AI Can Tell From Listening to You”.[i] McCormick said that by analyzing voice data, an AI can determine a person’s emotions, mental and physical health, height and weight, and detect if you are depressed, in danger of a heart attack, or dozing at the wheel of your car. The AI voice technology is already in use in a number of application areas. The following paragraphs contain some examples.
A major area of opportunity is in mental health. McCormick said,
In the U.S., mental illnesses affect one in five adults, or 46.6 million people in 2017, according to the National Institute of Mental Health, which estimates only half of those needing treatment receive it. Emerging voice technology may be able to make problems easier to spot.[ii]
CompanionMx Inc., a digital health technology company spawned out of the Massachusetts Institute of Technology, has introduced a mobile app, called Companion. Patients are encouraged to talk to the app describing how they are feeling. The app securely records the voice and extracts features which are indicative of digital biomarkers correlated with symptoms of mental health.[iii] Using AI, the relevant data and trends are available to a clinician who can make better clinical decisions to improve the mental health of the patient. Researchers who have studied the Companion system have found it very encouraging.[iv]
Another interesting area of opportunity is keeping drivers awake. McCormick said,
More than 800 Americans died falling asleep behind the wheel in 2015, according to October 2017 statistics from the National Highway Traffic Safety Administration, and more than 30,000 people were injured in crashes involving drowsy drivers.
Many cars already have voice recognition for making phone calls or telling the car GPS system where you want to go. Some cars have external cameras to help avoid collisions. Cars could also have a camera on the dashboard watching you. McCormick said,
Now, many major car companies and artificial-intelligence companies are designing AI that uses voice analysis, along with facial recognition, to assess the alertness and emotional state of a driver.
Toyota Motor Corporation displayed a demonstration vehicle at last year’s Consumer Electronics Show which can read facial expressions and voice tones. If onboard AI detects you are exhibiting signs of being tired, the car’s voice assistant could alert you and suggest you pull over to take a break. McCormick said that the AI in the car could engage you in a conversation and, over time, learn what topics of discussion are most likely to keep you alert.[v] It could also blast you with loud music to help keep you awake.
Other areas where McCormick reported AI and voice analysis will eventually have a large impact include:
Humanizing the call center
Hiring the Right Candidates
Stay tuned for more on these areas next week.
[i] John McCormick, “What AI Can Tell from Listening to You,” Wall Street Journal (2019), https://www.wsj.com/articles/what-ai-can-tell-from-listening-to-you-11554169408
[iii] “The Companion System Has Three Key Components,” CompanionMx (2019), https://companionmx.com/product/
[iv] McCormick, “What AI Can Tell from Listening to You”.