The answer to this question could come as a surprise. In the third part of our series on intelligent speech recognition, we deal with an area of voice analysis that is particularly important in interpersonal communication and interaction – and also in advertising.
“In communication, it is often the case that the emotional state of communication triggers a completely different subconscious trigger than just what you say”, explains Dagmar Schuller, CEO and co-founder of the Munich start-up audEERING. This is particularly true of politicians who use rhetorical stylistic devices, especially in election campaigns, which should not only convince the voters, but also set back their opponents. But how real and authentic do the candidates look in these situations? AudEERING conducted an analysis of Donald Trump and Hillary Clinton’s vocal personality profiles before the 2016 US election.
“We were extremely surprised that Trump showed a much higher approval in the voice than Hillary Clinton in the way he spoke, because that really is a neutral result. She had a much more negative voice profile than Trump,” says Schuller. Of course, one cannot say whether this had any influence on the outcome of the election, but the fact is that Trump has become president. “His voice seemed dominant and his credibility was extremely high, especially in comparison to Clinton.” Of course, there are people who do even better, but “in authenticity and honesty he was always the best.”
Austrian politicians are also “outed”
The company made a similar test with the top candidates in the National Council elections in Austria, Sebastian Kurz, Christian Kern and Heinz-Christian Strache. Using sensAI technology, the most concise emotional states such as happiness, enthusiasm, interest, disappointment and dejection, of the candidates were analyzed over the entire length of the only joint TV duel “The Big Three at Three”. The emotional activation (arousal/excitement/enthusiasm) and valence (pleasant/positive and unpleasant/negative perception) were also measured and aggregated in real time over the entire course of the TV show so that corresponding characteristics are recognizable.
“We found out, for example, that Sebastian Kurz was much, much more positive with his voice, his intonation, the way he spoke, than his two colleagues. With Strache we noticed that he was most emotional. At Kern, even though he didn’t say much negative, we noticed that the mood was permanently negative,” says Dagmar Schuller. “We heard a disappointment, almost a depressive impact. Of course, he may have that in his voice, but that’s bad for a politician.”
The way you talk is not only important in politics, but also in market research, as someone says. Just clicking Yes or No, Like or Don’t Like is far less meaningful than hearing someone say it. “Emotions always play a role in media and advertising, especially in the higher-priced section, what I buy or what I don’t buy. Advertising that has to be emotionally touching must be managed accordingly,” says Schuller. audEERING has done an analysis of the so-called influencers on YouTube to find out which videos are particularly popular, have the largest reach and the most positive comments and how they are presented. Through this, it was also possible to determine why one video had a bigger reach than another, regardless of who made it.
Different classes of emotions could indeed be observed, which apparently resulted in the same people’s videos being perceived as extremely bad, having a poorer reach and receiving negative comments. “That means there’s something that repels people or attracts them,” Schuller explains. “An example is that when you basically like someone, it is very often expressed through emotions. Then it is also important how I place a product, for example, with which influencer I want to create a positive connotation. Ideally, I want to pick up a customer who buys the product and is very satisfied and says it is the greatest thing he has ever bought. I don’t want to have a customer who was actually dissatisfied, made a frustration purchase and is even angrier afterward because the product cost so much and finally brings it back. I want to pick up the customer in a positive moment and make him even happier. Then I win.”
A prime example of this kind of positive advertising is Red Bull. For years, the energy drinks manufacturer has focused on communicating extremely emotionally and generating positive emotions with its potential customers. “Their aim is to establish a relationship between the top athlete and the fan who watches the top athlete and wants to put himself in his place. He wants to feel how the athlete feels when, for example, he jumps off the cliff or drives away from the avalanche. What adrenaline rush, what thrill has he got there?”
Dagmar Schuller emphasizes that this can also be used very well in the media sector of gaming. “Gaming has become something where players react very emotionally through role-playing or other things. If I can actually convey my feelings to my avatar in a certain combat situation, I have a completely new element that can be individualized”. On the other hand, such a mechanism also provides parents with an opportunity to better control their children if they notice that the offspring is too deeply absorbed in their game. “Then the system notices that the child gets too excited and gives the mother an alarm that warns her and recommends that she stops the game because it becomes unhealthy and too stressful for the child. You can do this very well with our software and it also has a very positive effect on a certain brand”.
Help for moderators
Another area of the software is voice coaching, for example for moderators, where the coach or even you yourself can get an analysis of your voice with the help of the software. “You can see how sympathetic someone else feels about a certain voice. That depends a lot on how my own voice sounds. The closer I am to my own voice in this perception, the more positive and pleasant the voice feels to me.”
In the area of voice coaching, you can train a voice specifically for the respective target group. “For children, a certain singsong might be a good thing, which is a little disturbing for adults. So you can say that the target group of your audience is adults, and if your voice doesn’t fit, you can train them accordingly. That is feasible,” emphasizes Schuller. “The voice coach can give feedback and say take it easy, take more breaks, you said that well, you said it less well.”
Such training can also be used for job interviews, she says. “If you train those interviews in advance, you get feedback from the coach, who says that you didn’t come across as convincingly now, do it again. I think there are still a lot of game options or variants in this area where you can tailor a basic software to the very specific problem. We can do that with a little adjustment.”
Perhaps one politician or another should give it a try…