RICHARDSON, TEXAS -
Call centers, it's where consumers go for product help and information, and often frustration. But UT Dallas professor John Hansen says he's trying to change that.
"What we would like to be able to do is identify in the audio stream where someone is unhappy."
Dr. Hansen is developing an algorithm that can do the math and detect anger in a callers voice. He says volume doesn't matter, but subtle voice traits change when people get mad.
"If you say help, you will probably change the way you produce the H or the P or maybe modulate the other sounds, the vowels and the consonants a little differently and those are the cues to kind of let someone know what your emotional state is."
Take this call for instance.
"How may I help you? The voice recognition system says. The caller responds. "I need help from a human being. I can't talk to a recording. This sucks, give me a @#%$&* human being."
George Platt is the executive vice president of Working Solutions inet. It's a call center support company that does business for nearly a hundred companies. Platt is joining with UT Dallas and will use the anger detection software. He says angry customers can be immediately redirected to supervisors who can take care of the problem.
"What we are hoping to do with this technology is to detect when customers are upset or angry and allow the company to react to those customers"
And, Platt adds, hopefully keep them as customers.
"And the goal here is that once we have those customers we don't want to lose them because they are upset."
Because angry customers often say this:
"No sir, I want to cancel now."
UT Dallas expects to roll out a prototype in a few months.
"What we would like to be able to do is identify in the audio stream where someone is unhappy."
Dr. Hansen is developing an algorithm that can do the math and detect anger in a callers voice. He says volume doesn't matter, but subtle voice traits change when people get mad.
"If you say help, you will probably change the way you produce the H or the P or maybe modulate the other sounds, the vowels and the consonants a little differently and those are the cues to kind of let someone know what your emotional state is."
Take this call for instance.
"How may I help you? The voice recognition system says. The caller responds. "I need help from a human being. I can't talk to a recording. This sucks, give me a @#%$&* human being."
George Platt is the executive vice president of Working Solutions inet. It's a call center support company that does business for nearly a hundred companies. Platt is joining with UT Dallas and will use the anger detection software. He says angry customers can be immediately redirected to supervisors who can take care of the problem.
"What we are hoping to do with this technology is to detect when customers are upset or angry and allow the company to react to those customers"
And, Platt adds, hopefully keep them as customers.
"And the goal here is that once we have those customers we don't want to lose them because they are upset."
Because angry customers often say this:
"No sir, I want to cancel now."
UT Dallas expects to roll out a prototype in a few months.
