Your Location: Home / News & Events / Media Resources / Speech Recognition - Giving Computers the Gift of Gab

	Related Information
	Feature Articles: Download PDF version of this article (PDF, 266 KB) Speech Technologies - Let Your Web Pages Do The Talking Products & Solutions: Advanced Speech Recognition Documentation: Web-Centric Self-Service service brief (PDF, 236 KB)

By Wendy Herman, Nortel

Speech recognition is coming of age, giving computers the ability to listen and reply in simple conversations that are bridging the language gap between people and machines to change the way business is conducted.

Improvements in speech recognition technologies have been steadily expanding the capabilities of computers to understand voice commands, making it possible to replace touch-tone keypad responses over the telephone with verbal ones. A wide range of businesses such as financial services, airlines, entertainment services, hotels, utilities and government agencies are implementing speech recognition telephony applications that complete repetitive tasks without needing to involve a human operator.

A customer can verbally transfer funds between bank accounts, make flight reservations or order concert tickets speaking only to a computer. Automating these types of transactions frees up a company's call center staff to focus on more complex customer needs while contributing large savings to customer service budgets. An industry analyst, Zelos group, estimates a traditional call center can reduce the cost per customer call from $4 to $10 to between 10 and 40 cents through automation. Customer satisfaction also improves through reduced waiting times.

Speech recognition is listed as one of the top 10 technologies likely to have the greatest impact on business through to 2007, according to the Gartner Group, a leading provider of research and analysis on the global information technology industry. An Allied Business Intelligence study estimates the speech recognition market could grow to $53 billion globally by 2008 from its current $1 billion.

"Over the past few years, speech recognition applications have improved to include large complex vocabularies that go well beyond simple commands like yes or no and the newest speech synthesizers can sound just like a real person talking rather than a machine," says Tom Neary, who leads a voice product development team for Nortel Enterprise Multimedia Applications.

Nortel, an industry leader in speech technologies for more than 25 years, received the Frost & Sullivan 2004 Speech Solutions Competitive Strategy Award for its accomplishments in deploying speech technologies for enterprises. Frost & Sullivan also selected Nortel as the recipient of its 2004 Market Leadership Award for the company's lead in the U.S. Interactive Voice Response (IVR) market.

Nortel has designed and deployed over 200 speech applications in more than 16 countries around the world, in 15 languages, according to Frost and Sullivan market research. The company introduced the world's first large-scale voice recognition application for telephone in 1989 with its Automated Alternate Billing Service using simple yes or no prompts to navigate a menu. In 1992, it introduced the world's first large-vocabulary dial-in speech recognition service Stock Talk for the financial services industry.

At SpeechTEK 2004, the world's largest speech technology conference held last year in New York City. Nortel's Manish Sharma was named one of the 10 Top Leader's in Speech for his contributions to the speech technology industry over the past year. Sharma leads a Nortel team of speech applications designers and developers that customizes products to meet the specific needs of individual customers. Nortel's IVR platform's technology uses software from world leaders in speech technologies - Nuance* and ScanSoft*. The company develops a variety of products in-house to run on the platform as well as customized solutions.

Nortel Networks Interactive Voice Response (IVR) platform's technology uses software from world leaders in speech technologies - Nuance* and ScanSoft*. The company develops a variety of products in-house to run on the platform as well as customized solutions to meet specific customer needs.

Part of the growing appeal of speech recognition applications, Neary says, is they have become increasingly intuitive and faster to use than keypad responses. Instead of giving one word responses for each step through a long menu of questions, he says, "you can speak a sentence and the application will pick out multiple key words to select the appropriate response and complete the transaction much more quickly than navigating a menu."

"In banking for example, once you've given your account number, password and the system has verified your voice against your voice print on file you can say -- I'd like to transfer $100 from my chequing to my savings account. You're then given a transaction number and you're done. It's much faster than punching a series of commands and numbers on the telephone key pad."

One Nortel customer, Lloyds TSB with bank branches throughout the UK uses advanced speech technology to handle millions of calls a week. When Lloyds first introduced telephone banking five years ago with call center agents, it found customers reluctant to discuss their personal financial details over the phone. Once Lloyd's implemented an automated teller that enabled customers to examine their accounts with privacy, use of its telephone banking service took off. By November 2003, it had received its 70 millionth call with account holders contacting the bank on average three times a month.

A Nortel health services customer in the U.S. which has two million customers is using speech recognition to relieve a patient backlog problem where confirmation of appointments requires patients and physicians to make repeat calls. The company was able to move more than 30 different telephone and fax numbers to one 800 number. Appointments are now set up and confirmed through one quick call.

Besides call center applications, Nortel is introducing speech recognition within its Web-Centric Self-Service (WCSS), giving people the option to access web site information on the computer or listen to it by telephone. WCSS uses industry standards such as Voice eXtensible Markup Language (VXML) and Call Control eXtensible Markup Language (CCXML) to give voice recognition capabilities to web applications. In the same way the HTML standard brings a common formatting for putting text on the web, VoiceXML formats voice.

"Bringing voice to web applications to access content by telephone gives people real-time access to information whenever they need it, regardless of location," says Tricia Schneider, speech applications marketing prime for Nortel's Enterprise Multimedia Applications.

For example, if someone is driving and wants directions to the nearest Mexican restaurant, Schneider says, they can call a map service web site, describe where they want to go then listen to directions on the telephone. The big advantage of WCSS for a company's web pages, she says, is that they don't have to re-design the applications on their site to make them accessible by telephone. The two industry standards for formatting - HTML and XML - work together. Text can easily be transferred to spoken information through WCSS.

The applications for use of text to speech extends beyond call centers to the virtual office, adding another element to unified messaging which can now collect all telephone, email and fax messages onto one convenient inbox, the same as what's currently used for email. With speech recognition and text to speech added, written email or fax messages can be listened to over the telephone when the recipient may not have access to a computer or PDA.

"As with any new technology, its widespread use grows as the consumer's comfort level with it increases," says Neary. "Over the next two years, speech recognition will become more and more commonplace until, like so many applications, we'll wonder what we ever did without it."

Get the plug-in*