Whether it is the story of Peter Piper and his pickled peppers or a woman selling sea shells on the seashore, tongue-twisters tackled when sober can sound rather different after a drink.
Now researchers believe such changes, in particular those relating to pitch and frequency, could be used to alert people to their level of intoxication.
Dr Brian Suffoletto, the first author of a study from Stanford University, said the approach has a number of potential future applications. “The most obvious one is as a form of ignition lock on cars which would not allow someone to start their car unless they could pass the ‘voice challenge’ which could be used in certain high-risk workplaces like school bus driver or heavy machine operator to ensure public safety.
“Another application could be in restaurants or bars so a bartender can know when to cut someone off.”
Writing in the Journal of Studies on Alcohol and Drugs, Suffoletto and colleagues report how 18 adults aged 21 and over were initially recorded reciting a tongue-twister.
Each participant was then given a large, weight-based dose of alcohol – enough to intoxicate them – and subsequently asked to recite a different tongue-twister each hour, up to seven hours after consuming the alcohol. This speech was recorded.
Suffoletto said: “The tongue-twister was used as a kind of vocal stress test to bring out changes that might not be detected when speaking routine prose.”
The researchers also took breath alcohol readings from participants before they consumed the drink, and every 30 minutes afterwards until the seven-hour post-drink point.
The recordings of the rhymes were split into one-second windows and the features of the voice relating to pitch and frequency were analysed. An AI system was then trained on a subset of this data with the corresponding breath alcohol concentrations, before it was tested using the remaining voice data.
The results suggest the voice analysis predicted alcohol intoxication with 98% accuracy, with intoxication defined as a breath alcohol concentration equal of more than 0.08% – the legal blood alcohol limit in the US for driving.
The study has limitations, however, including that it only included white participants and did not look at other features of speech that may be influenced by alcohol such as its volume.
Suffoletto added that use of harm prevention messaging also requires careful timing. “For instance, as someone initiates drinking, a reminder of their consumption limits can be impactful. However, once they are significantly intoxicated, the efficacy of such interventions diminishes,” he said.
Petra Meier, professor of public health at the University of Glasgow, who was not involved in the research, stressed the study was small and had tightly controlled conditions.
“I believe that there is the potential for exciting developments that could eventually be really useful, but obviously one would first want to test this approach in larger and more diverse samples,” she said, adding it would then be important to test the approach in real-life situations.
Meier said: “For example [to] check that it works equally well for light drinkers and for regular, heavy drinkers who may not [show vocal changes] until they reach much higher levels of intoxication and might be wrongly reassured they are fine to drive.”
