Many kiosks and self-service devices utilize a human voice in addition to on-screen prompts - and most often, that voice is female. NCR lab tests show why this is effective, and how it translates in other countries.
April 16, 2006 by James Bickers — Editor, Networld Alliance
If you close your eyes and think back to the last time you used a self-checkout lane at a grocery store, chances are you can conjure up the machine's voice:
"Please scan your first item."
"Please place the item in the bag."
"Do you have any coupons?"
While not all self-service devices use a voice guide, self-checkout generally does. And generally, it is a female voice - and from a real woman.
"We have found from user acceptance testing in our labs and in field studies that people, particularly in retail settings, respond better to a female voice that is not synthesized," said Marcia Crosland, manager of user experience consulting with NCR. "Synthesized voice for the type of self-service that we're using is not quite advanced enough in order to be able to sound natural."
Crosland said voice is used on self-checkout for two reasons - to help communicate instructional information, and to make the user feel more comfortable with the machine. "We want it to feel as natural as possible," she said. "Granted, it's not the same as a human-to-human interaction, but we want the human-to-machine interaction to be as natural as possible, and be very pleasing to the user."
In the early days of NCR's self-checkout system, both male and female voices were used. Crosland said the female voices were utilized on the routine transaction functions - the welcome, the "thank you" - while the male voice was reserved for security. If you did something wrong or inappropriate, a male voice would pipe up and tell you, in a very stern tone, to take corrective action.
"While that was effective in catching the attention of the consumer, it was not very pleasing," she said. "We were trying to have a wider acceptance for self-checkout, so the last thing we wanted to do was to offend the user."
The company performed a series of lab trials, combined the data with real-world observation, and found that customers overwhelmingly responded better to the female voice. The general population was more accepting of a female voice giving them general instruction.
A voice that knows no boundaries
Since voices are so closely tied to the region from which they emanate, one would expect geography to play a big role in what gender voices work for self-service. Surprisingly, that hasn't been the case - female voices seem to work equally well in any country.
"We're still finding, contrary to opinion, that the female voice is highly accepted," Crosland said. "When we were selecting voices in Germany and Turkey and Russia, there was the thinking that perhaps the voice should be male. But we found just the opposite. In patriarchal societies, sometimes the males are more accepting of a female voice telling them to take an action."
Crosland said NCR hires and maintains vocal talent in each of the 15 countries where its self-checkout solution is employed. In addition, regional dialects are reflected. "For example, in Spain we have a Castilian voice, and we found that she has a beautiful dialect that would not be accepted in Mexico - we have a separate voice in Mexico. Now, we're being asked in Spain to add two new dialects that are regional, because there are variations in the way we speak. And we want the machine to feel as if it's a part of your environment, and very natural."
NCR's findings on user acceptance serve as a reminder to all interface designers: Don't assume you know your audience. You cannot simply listen to what they tell you - you must also watch them in the real world, to see how they really behave.
"The psychology of the user is often a study of contradictions," Crosland said. "What you think is intuitive and seems very easy on the surface is really the opposite. That's why engineering technology is a science and not an art."