Talk back to the robo-voice that answers Kootenai Electric Cooperatives telephone, and it might help you do what you want instead of rambling on about buttons you can push on your phone to reach your party.


The Hayden-based power providers telephone system has speech recognition. In addition to being able to access the typical automatic-attendant instructions that have become commonplaceTo report a power outage, press one. For customer service, press three.a caller can say an employees name, and the system will route the call to his or her desk.


Auto-attendant, as a general rule, is not well-liked, says Chris Cheeley, president of Coeur dAlene-based Phones Plus Inc. The speech-recognition systems are faster, more accurate, and more customer pleasing.


Such systems still are on the leading edge of technology, Cheeley says, but are on their way to becoming more mainstream.


Chanley Geveshausen, a regional manager for Spokane-based communications-systems provider Cerium Networks LLC, says about 70 percent of the automatic-answering telephone systems Cerium sells now include a speech-recognition component. In general, he says, sales of such systems are growing in popularity.


Its a do-more-with-less economy, Geveshausen says. Theres a huge return on investment, because information can be self-service.


Speech-recognition components can be added to a phone system or installed with a new system. The speech-recognition component alone costs at least $15,000, Cheeley says, and can cost tens of thousands of dollars more than that, depending on the size of the phone system, he and Geveshausen both say.


Cheeley says most customers who have such a component have at least 50 employees and usually many more than that. Geveshausen says, however, Cerium Networks has installed such components for smaller companies, in addition to its large clients.


Cheeley says customers often will use the speech-recognition technology internally for a while before using it to answer incoming calls. For instance, Kootenai County, which employs about 750 people, had installed a new telephone system earlier this year and is using speech recognition only for internal calling and call transferring. Now, if a county commissioner wants to call a building inspector, he can pick up the phone and simply say the inspectors name, rather than having to look through the county directory to get the inspectors number.


Incoming calls to Kootenai County still are greeted by a conventional automatic attendant, so callers arent able to use the speech-recognition technology. Cheeley says he doesnt know whether the county plans to begin using speech recognition to direct callers. Because the countrys residents vary greatly in age and technological savvy, he would caution the county to craft its greeting carefully to avoid confusing people who are unaccustomed to such technology.


Even if the county doesnt use speech recognition for outside callers, the technology saves the county money, Cheeley says, because it doesnt have to print new telephone directories for its staff every time theres turnover. Previously, he says, the county printed new telephone directories for its employees at least monthly.


Kootenai Electric, on the other hand, uses speech recognition both internally and to greet incoming calls. Art Malin, Kootenai Electrics systems administrator, says the system will ring an employees desk and cell phone at the same time to ensure that the employee gets the call.


Theres a lot less phone tag that goes on with this system, Malin says.


In addition to improving communications, the new phone system with speech recognition has saved the electric cooperative from eventually having to hire additional front-desk help, Malin says. Before the new phone system was installed, one front-desk receptionist fielded telephone calls full time, and the other helped customers who came into the office. Walk-in traffic had increased to the point Kootenai Electric was debating whether to hire a third person to help out up front. Instead, since the new phone system handles most incoming calls, both receptionists are free to help customers in person, and a third person no longer is needed.


Kootenai Electrics telephone system is linked to the cooperatives e-mail server, as are many similar systems with speech recognition. This provides a couple of advantages, Malin says.


In addition to being able to use speech recognition to call people in the database, employees can call in and use the technology to call contacts in their own Microsoft Outlook file. Instead of looking up a contacts phone number, an employee simply can say the persons name, and his or her number will be dialed. Malin says this feature especially is handy for employees who are in the field and dont have their contacts at their fingertips.


Also, Kootenai Electrics phone system includes a text-to-voice feature, so that an employee can have the speech-recognition system read his or her e-mail messages aloud. An employee can respond to an e-mail with a voice message, which is sent as an attached, sound-bite like .wav file to the e-mail reply.


At this stage, text-to-speech technology is prevalent, but speech-to-text technology either is incredibly expensive or lousy, Cheeley says. In general, he says, speech-to-text is more difficult to pull off because people often talk imperfectly, saying um or ah a lot. Such sounds can confuse a computer as it tries to decipher words and translate them into text.


Of course, in a text-to-speech system, pronunciation recognition for some names and words typically needs to be added to the systems database, which can be done easily, Cheeley says. Such a database already contains common, nonphonetic exceptionsit knows, for example, that science isnt pronounced sky-ence. It might not know, however, that Julio is pronounced hoo-lee-oh rather than jew-lee-oh.


Geveshausen says, however, that multilingual speech-recognition components are available in both different languages and different dialects. So, a company that does business in Canada might have U.S. English and Canadian French installed on its system, along with Canadian English. A company that does business in England might set up its system to recognize both U.S. English and British English, so that it would acknowledge the difference in speech.


In coming years, additional applications of speech-enabled technology likely will become prevalent, Cheeley says. Speech-recognition technology might be used commonly in computers, for example, so that a user could open a file or start an e-mail with a voice command.