Researchers from Stanford and Toronto Universities have developed an algorithm, integrated into a mobile app, capable of detecting alcohol intoxication from the user’s voice recording. This innovation could revolutionize how we measure alcohol in the body and contribute to preventing alcohol-related road accidents.
A New Approach to Measuring Alcohol
Despite numerous awareness campaigns, driving under the influence of alcohol remains a major cause of road accidents. Breathalyzers, although very useful for roadside checks, are expensive and impractical. It is in this context that researchers have developed an algorithm for mobile phones capable of calculating the level of alcohol intoxication by simply recording the user’s voice.
The idea behind this approach is simple: almost everyone has a mobile phone today, so it would be enough to install the application. Moreover, alcohol could be measured regularly without the user having to report it. The phone would record the user’s voice in daily conversations at regular intervals and if dangerous intoxication is detected, an alarm would be triggered.
How Does It Work?
Slurred speech and a tangled tongue are among the first signs of alcohol intoxication. But there may also be vocal parameters imperceptible to us but common to most drunkenness. This is where artificial intelligence comes in. The researchers used machine learning to design an algorithm capable of detecting vocal patterns that may be linked to alcoholic intoxication.
To train this algorithm, they used 18 volunteers over 21 years old who were given a controlled dose of alcohol. They were then asked to read aloud a difficult sentence while their breath alcohol level was measured. This procedure was repeated for seven hours, allowing the algorithm to detect changes in the voice based on the amount of alcohol in the exhaled breath. The results are promising: once trained, the artificial intelligence achieved a 98% accuracy in measuring alcohol from vocal recordings.
Challenges to Overcome
However, this study remains limited and researchers acknowledge they would like to conduct more experiments with more volunteers of different ethnicities. Some people tolerate alcohol less well than others due to genetic factors. Moreover, the application should be able to operate automatically and take regular measurements. If the user is in a crowded bar or chatting with other people, this could complicate detection of changes in speech.
In addition, if the user has drunk and is heading alone to his car, other parameters should be taken into account, such as walking style. If the user stays seated at the bar, the algorithm could also analyze how he drafts text messages. The goal is to have an algorithm capable of detecting anomalies in numerous parameters and resorting to each of them or a combination thereof as needed.
A Promising Future
There’s still a long way to go for measuring alcohol with these types of devices. But the results obtained so far are encouraging and this technology could well be part of our future. In the meantime, it is important to remember that if you drink, you should not drive. This application could be a valuable tool in helping to prevent alcohol-related road accidents.
I am a web editor specialized in people news. I am passionate about the world of celebrity and I love to follow the latest trends, scoops and gossip that make the buzz