Rather than formally gather information about customers' level of satisfaction through user surveys or through speech recognition, converting speech to text and then analysing what's being said, the novel technology takes into account the average pitch of the voice, its degree of variation, and also characteristic changes at relative points within voice data that covers multiple words, such as the start or end of speech.
A customer's "voice cheerfulness" emerges from patterns of changes in voice pitch, usually at a high tone or when the voice's tone and volume change a great deal, explain the researchers. From these changes of pitch, proprietary conversion algorithms are able to identify the unique characteristics of a cheerful voice, especially at the beginnings and endings of conversational statements. The perceived voice cheerfulness correlates well to the degree of customer satisfaction, Fujitsu Laboratories claims. Combining this with customer-service evaluations, and using machine learning, a threshold point between satisfaction and dissatisfaction can be set, meaning that just listening in, AI software could identify on-the-fly when a customer is satisfied or dissatisfied.
Users of such technology could adopt their marketing or conversational tactics on the fly too.