The Art and Business of Speech Recognition: Creating the Noble Voice

Have you ever heard a radio commercial for a car dealer in which the announcer speed-reads the financing terms and conditions in the final five seconds? Like the fine print in a newspaper ad, that information has to be there by law ”never mind that no listener can actually understand what's being said!

Many speech-recognition applications also require legal disclaimers or notices to be read to callers at some point ”usually to protect the client from liability. Legal language needs to be precise, but a good lawyer should be able to make the statement simple and clear without using any confusing or arcane legal terminology. After all, the point is to have lay people understand the terms and conditions of the application. A lawyer will also be able to tell the designer the conditions under which the disclaimer must be played , so that the designer can incorporate this text at the right points in the application and can determine if a disclaimer may or may not allow the caller to interrupt the prompt.

The downside of including a legal notice in an application is that it can be disruptive, user -unfriendly, and time-consuming . I found that out for myself not long ago when I signed up for voice mail from my local telephone company. Immediately after I dialed in for the first time, the system launched ”actually, exploded is a better word ”into a stream of poorly written legal text that ran about two minutes. I was required to listen to the whole thing and then press "1" to accept the terms and conditions before I could even learn about my voice mail account.

Could this have been handled in a more palatable way? Yes. The voice on the system should have first introduced the subject by saying that it was about to read a required legal statement about the terms and conditions of using the voice mail system. It should have prefaced the statement by summarizing its purpose in simple language. And it should have informed me about how long it would take. A better approach might be:

"OK, before we get going I have to read the official terms and conditions of using this system. Basically, it will tell you about how [the phone company] isn't liable in the event that technical failures make it impossible for us to retrieve particular voice messages. If you want to learn more about it after I read it, you can press 'zero' and talk to an operator. OK, here goes ”it takes about a minute and a half: 'Whereas the provider of the voice mail service <rest of the legal statement>.'"

After playing the statement, the system should ask callers if they want to hear it again (not that anyone would) or tell them how to accept the terms. The big advantage of this technique is that it doesn't disrupt the collaborative relationship the system is trying to develop with the caller. It simply excuses itself for having to read a statement that is totally inconsistent with the rest of its personality.