Our Terms & Conditions | Our Privacy Policy
Voice AI agents are growing in popularity.
getty
As businesses transform their contact centers to include AI agents, many are struggling with a complex question: What should those agents sound like?
Businesses have more options than they could ever consider. There are voices based on real-life voice actors, whose recordings can be used to have an AI tool say just about anything. And there are synthetic, entirely AI voices. In the past, those may have been easier to distinguish. But these days, some AI voices sound just as natural as real ones.
Digging into this question can seem daunting — not only because of the array of possibilities, but also because it involves lingo most people have never had to think about. Here’s a guide for you to make sense of this, and simple ways to help guarantee success.
From ‘sociophenetics’ to ‘illocutions’
Just weeks ago, a group of researchers in the United States and China published a study titled AI Voice Matters: AI’s Pragmatics and Sociophonetics in Persuasion. They experimented to determine how various AI voices affected potential customers in phone calls. Their conclusions show the many moving pieces to consider.
For example, they looked at “coquetry,” a term for a playful, often flirtatious style in how an AI voice can be designed. They found that it didn’t help. In fact, the opposite occurred. “We observed a direct negative effect of a coquetry voice on persuasion, particularly for male chatbots,” the team from the Beijing University of Technology, the University of Texas at Austin, and the University of Memphis reported. Meanwhile, “The stern voice was found to be similarly effective as the gentle voice. Moreover, a male stern voice prompted customers to ask more questions, thereby increasing its persuasive power.”
How about age? “There was no significant difference between the persuasive effects of young and old voices, the group (Lin Jiaa, Xiayu Hub, Yifan Yuc, and Huigang Liang) wrote. “Additionally, an old voice’s persuasion effect is partly through extended conversation duration” — yet another element to consider.
They also examined the type of language structure these voices used, and whether customers responded differently when various structures were use by different voices. “An affirmative sentence, particularly in female voices, led to an increase in customer inquiries and in turn enhanced the overall persuasive effect.” Rhetorical sentences and questions were less effective, the study added, referring to all of these as “illocutionary acts.”
Ultimately, the right voices for your organization will also depend on factors specific to your brand.
Why a unified model is essential
Because my work at Nextiva involves helping organizations transform their CX (customer experience), I go through this process with clients. I explain to them that, in assessing which ones work best, they need to keep in mind the multiple end goals. Sure, ideally each interaction will lead to a purchase or a resolved issue. But you also want each interaction to leave the customer with a lasting, positive impression of the brand, even if those other goals were not achieved.
Testing out voices requires a comprehensive system. You need to get real-time information on how customers respond in multiple scenarios. Ultimately, you may arrive at a point of extreme personalization, in which specific voices automatically come up for each customer based on their history.
For all this to happen, your company needs to gather all of the information it can about customers into a single record. That’s why unified customer experience management is vital. An AI-powered UCXM platform consolidates customer interactions, gets rid of silos and provides a holistic view of the customer. Your entire company operates within the same platform, so any new information relevant to each customer pops up immediately (with privacy and compliance covered of course). The tools highlight key insights at a glance.
In this era, that kind of omnichannel communication system is a necessity. A UCXM quickly helps brands develop their “voices.” That used to be a figurative term referring to the kinds of language organizations use; now, it’s become literal as well.
As you test out voices to build into your UCXM, you’ll most likely experience some stumbles. That’s OK. With technology advancing so quickly, there are always new hurdles to overcome. But you’ll learn from them, and keep moving forward. After all, every fall is a chance to get back up and grow stronger.
Images are for reference only.Images and contents gathered automatic from google or 3rd party sources.All rights on the images and contents are with their legal original owners.
Aggregated From –
Comments are closed.