If we consider the tale of Peter Piper and his preserved peppers or a lady vending sea shells at the beach, reciting challenging phrases may sound distinct after consuming alcohol instead of being sober.

Currently, scientists theorize that modifications in vocal pitch and frequency could potentially serve as indicators of one’s level of intoxication.

According to Dr. Brian Suffoletto, the lead researcher at Stanford University, the method has numerous potential uses in the future. One such use could be as an ignition lock for cars, preventing individuals from starting their vehicle unless they can successfully complete a “voice challenge.” This could also be implemented in high-risk work environments, such as for school bus drivers or heavy machinery operators, to ensure the safety of the public.

Another possible use for this technology could be in restaurants or bars, where a bartender could receive notifications when it is necessary to stop serving alcohol to a customer.

In a study published in the Journal of Studies on Alcohol and Drugs, Suffoletto and his team share findings on 18 participants over the age of 21 who were observed performing a tongue-twister.

Each person was provided with a significant amount of alcohol based on their weight, enough to make them intoxicated. They were then instructed to say a different tongue-twister every hour for up to seven hours after drinking the alcohol. The recordings of their speech were saved.

Suffoletto stated that the tongue-twister was utilized as a vocal stress test to reveal potential changes that may not be noticeable when speaking in a normal manner.

The scientists also recorded participants’ breath alcohol levels before and every half hour after they drank until seven hours had passed.

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 findings indicate that voice analysis accurately predicted alcohol intoxication at a rate of 98%, using a breath alcohol concentration of 0.08% or higher as the legal limit for driving in the US.

However, the study does have limitations. It only involved participants who were white and did not examine other aspects of speech that could be affected by alcohol, such as its loudness.

Suffoletto mentioned that the effectiveness of harm prevention messaging depends on timing. For example, a reminder of consumption limits may be impactful as someone starts drinking. However, once they are heavily intoxicated, the effectiveness of such interventions decreases.

According to Petra Meier, a public health professor at the University of Glasgow who was not part of the study, the research was limited in size and carefully monitored.

She stated that there is potential for exciting advancements that could potentially be highly beneficial. However, it would be necessary to first test this method on larger and more diverse groups before implementing it in practical scenarios.

Meier stated that it is important to ensure that a particular method is effective for both light and regular, heavy drinkers. This is because heavy drinkers may not exhibit vocal changes until they have reached a higher level of intoxication, potentially leading them to believe they are capable of driving when they are not.

Source: theguardian.com