We are often told that you only get one chance to make a first impression and for years people have questioned just how to go about it.

But it can be even harder to create the right first impression when it is only your voice you are being judged on.

Now scientists at the Institute of Neuroscience and Psychology, University of Glasgow and at the La Timone Neuroscience Institute, Aix-Marseille University, France, have found a way to modulate a voice pattern which a listener perceive as “trustworthy”. 

They believe that this could open the door to a new synthetic voice personality that would give the best first impression to listeners.

And the scientists think this could have applications from everything from your car’s SatNav to automated telephone banking. It could also be used to help those people who have lost their voices through disability or injury by creating new artificial voice boxes with personality.

The Sound of Trustworthiness research article by Dr Phil McAleer of the University of Glasgow and Professor Pascal Belin of Aix-Marseille University is published in the Public Library of Science (PLOS).

Dr Phil McAleer, a University of Glasgow lecturer in psychology, said: “We were really surprised by how faithfully the personality impressions followed our acoustical manipulation. It was almost like we had a sliding scale of trustworthiness.

“The more monotone or flat the voice sounded, the less appealing it was to the listener. While a voice with a bit more personality and intonation gave the best and most positive impression. It was as if we were turning a ‘voice trustworthiness’ button on a sound system.”

Professor Pascal Belin, the lead author of the study, said: ““These results demonstrate that there is a strong acoustical basis to our voice personality impression.

“We have now a principled method to modulate the personality impressions that voices elicit. Our results were obtained with trustworthiness but there is no reason why it should not work with other personality traits.

“Thus we have now a powerful way to generate voices eliciting the desired personality impressions, with many potential “vocal make-up” applications in modified or artificial voices”.

Using computer modelling and web-based voice judgments, the scientists were able to generate voices to gauge how trustworthy they appear to a listener.

Research shows that when we hear a new voice, we form personality impressions, such as trustworthiness or dominance. These “first impressions” are quite consistent across listeners, even in a single word like “hello”.

But the acoustical basis for this phenomenon has remained elusive: how exactly should one say “hello” to be perceived as trustworthy by new listeners? The scientists have now solved that problem.

They first generated two synthetic voice prototypes by averaging recordings from individuals rated high or low in perceived trustworthiness. Then they generated a continuous series of voices between these two prototypes (listen here), which they asked 500 listeners to rate on trustworthiness via a web interface.

Their article says the results have been quite spectacular: the trustworthiness ratings increased in a near-straight line by moving along the voice continuum.

To put it simply the more monotone a voice sounds the less trustworthy it is perceived, while a more “sing song” one with personality and intonation gives the listener a good impression of the speaker.





First published: 13 October 2017

<< October