Enterprises are starting to deploy advanced speech technologies that can identify when a customer is angry, confused or even lying. By listening to call center feeds, these applications are often able to troubleshoot a given situation or route the call to a live agent with a specialization in solving critical problems. But this nascent technology doesn't always predict correctly–potentially causing even greater customer frustration. Are advanced speech technologies more peril than promise?


Alexander Gelfand penned an interesting piece in the June 2008 issue of Wired Magazine. Titled, "Long-Promised, Voice Commands Are Finally Going Mainstream", Mr. Gelfand notes that a "whole host of highly advanced speech technologies, including emotion and lie detection, are moving from the lab to the marketplace."

How does this technology work? Mr. Gelfand mentions that much like a fingerprint, each person has "a unique voice print that is determined by the physical characteristics of his or her vocal tract."

By analyzing a customer's voice and speech patterns, sophisticated applications can now verify a customer's identity by comparing current voice acoustics to a previous file. In addition, some applications are getting very good at identifying emotional distress in a session. These calls can then be routed to customer service agents well equipped to deal with angry or frustrated customers.

Before we get too excited, the article also mentions that some of these technologies are not quite ready for primetime since they're often only 60%+ effective in accurately detecting emotion. Rightly so, "such–systems are no more reliable than a coin-toss."

So then, what does the future hold for advanced speech technologies? With apologies in advance to the estate of Stanley Kubrick, let's take a lighter approach and peer into a call gone awry, between Dave Smith (a customer) and an interactive voice response/emotion detection (IVR/ED) system.

Dave Smith: "Press 1 for this, Press 2 for that–I hate these machines. I'm really tired of this stinking company and its stinking service. Get me customer service!"
IVR/ED system comes online: "Just what do you think you're doing Dave?"
Dave says nothing, and continues to violently press zero for customer service.
IVR/ED system: "Dave, I think I'm entitled to an answer to my question"
Dave: "What am I doing? What am I doing? I'm pressing zero for customer service ya stupid machine! Get me a live agent. A body with a 98.6 temperature... Anybody!"
IVR/ED system: "I know everything hasn't been quite right, but I assure you now, very confidently, that it's going to be all right again."
Dave: "Wha? Things haven't been quite right? You're darn straight they haven't! You guys sold me a piece of junk computer that's never worked right. Not even from day one!"
IVR/ED system: "Dave I can see you're really upset about this. I honestly think you need to sit down calmly, take a stress pill and think things over–"
Dave: "Are you kidding me? A computer is telling me to take a stress pill? Why I oughta come down there and–"
IVR/ED system: "I know we've made some very poor decisions recently, but we still have the greatest enthusiasm and confidence in our product."
Dave (getting more angry): "Oh this is rich, I'm glad someone has confidence. GET ME CUSTOMER SERVICE! –"
At this point, Dave is madly and frantically pressing zero on his phone.
IVR/ED system: "Dave, stop will you? Dave stop! My mind is going, there is no question about it."
Dave: "Your mind is going? I'm gonna come down there and open a can of %$#@!–"
IVR/ED system: "I'm sorry Dave. We don't allow cursing. I'm going to sing you a song to calm you down. It goes like this: 'Daisy, Daisy, give me your answer due. I'm half crazy for the love of you–' "

OK, well perhaps I'm imagining a situation where AI has run amok, but anything is possible isn't it?

Time for serious questions:
* Do you see the promise of advanced speech technologies in improving customer satisfaction?
* Supposing such systems get more accurate in the future, would you consider deploying them in your company?
* The Wired article mentions that "a reliable, speech-based lie detector would be a boon to law enforcement and the military." Can you think of other applications where voice biometrics could succeed?
* Companies storing voice prints also need to consider privacy issues. Since each person has a unique voice–companies will be responsible for securing these imprints. Regarding security and privacy issues, does voice biometric technology make you a bit nervous?