Posted
Feb 1, 2006
 | By

We're talking. Is anyone or anything listening?

The dream for many is to talk into a microphone attached to your humble PC and have it produce a perfectly formatted document. The reality for an increasing number of people is for their spoken words to trigger a series of steps in rather more mundane applications.

The experience of BOC highlights this. The company is a major provider of gases and related products. These are sold through a network of more than 900 sales agents in Australia and across the South Pacific.

The process of recording sales, generating invoices and associated transactions involved an increasingly complex paper trail. Ben Molenaar, eBusiness manager for BOC, said the more the company looked at the problem, the more it looked like an application for a speech recognition system.

In the past, agents recorded sales on a paper-based system and mailed them to BOC. The new system, named AVRIL (Automated Voice Response Information Line), was developed to automate account status, sales transactions and credit card payments.

BOC sales agents now make a toll-free phone call to AVRIL and once a PIN is entered, callers can then speak to AVRIL directly. The system asks for the information required and then the details are recorded and automatically processed by BOC's host system. At the end of the call, a simple "that's all" or "I'm finished" ends the conversation.

Molenaar said the new system has increased the number of calls that can be serviced over the phone therefore eliminating long hold times and call queuing. "Other than entering their PIN, callers do not have to struggle with confusing push-button menus because they can simply speak to AVRIL in the same way that they would speak to a call centre agent," he said.

"It means the sales are recorded on the day they are made," Ben Molenaar said. "But for the system to be successful we had to make it simple to use. We also had to answer the question, 'What's in it for me?' coming from our sales agents.

"What's in it for them is a quicker and simpler system with fewer errors and more accurate billing.

"The key to the success of any speech recognition system is it has to be easy to use and encourage the user to use it frequently," Molenaar said. AVRIL was developed for BOC by VeCommerce.

Paul Magee, VeCommerce's managing director, said it was important to understand the critical nature of each customer's call centre experience. "Bad customer service, be it from an automated system or a 'traditional' call centre agent, is always unacceptable," he said.

"BOC's decision to provide a faster and more efficient service where the customer can be heard is to be applauded. AVRIL will deliver more control over transactions, provide accurate customer records and up-to-date information, and ultimately provide a higher level of customer service."

While appearing compelling, speech recognition has seen two of the largest companies in the IT industry - IBM and Microsoft - treading very slowly, if at all, in speech recognition technology.

Microsoft's Speech Server 2004 has yet to set the world on fire but there are positive signs the company is pushing ahead with the technology. It recently acquired intellectual property assets from Unveil Technologies, a developer of enterprise call centre technology.

Microsoft plans to incorporate the Adaptive Learning and Conversation Assist components of the Unveil Conversation Suite into future versions of Microsoft Speech Server. It says this supports its continued commitment to deliver the most flexible and integrated speech platform at the lowest total cost of ownership.

In response to requests for information for this article, Microsoft only offered a speech delivered more than 12 months ago by Bill Gates.

IBM was even less responsive. The company has given up support for its Via Voice product which is now handled by Nuance Communications. This gives Nuance Communications a leading position in the market for speech recognition applications. Even Nuance Communications recognises this will not last forever.

Peter Chidiac, Asia Pacific regional director for Speechworks (a division of Nuance Communications), says IBM and Microsoft have found speech recognition to be more challenging than they expected.

"They will catch up," he said. Nuance Communications now enjoys better than 70 per cent share of the market for speech recognition systems.

As for the major applications, Chidiac says the obvious ones are applications within call centres. "This will free up agents in call centres to handle higher value transactions," he said.

While speech technology has not suffered from a lack of hype, many companies are dragging their feet when it comes to even experimenting with speech. According to Benchmark Portal in the US, these companies may have reached the end of their dithering and are ready to make a commitment: according to its research, 50 per cent of companies currently not using speech are considering implementing it within the next year.

One of the inhibiting factors locally has been the well-publicised but less-than-successful speech recognition systems. One was for the directory service of a large telco and the other a booking service for a taxi company.

"These systems certainly had a negative impact on speech recognition systems but I think we're over that now," says Chidiac.

Indeed, one telco has been recognised for the success of its speech recognition system. Vodafone's Lara system is used to register pre-paid customers of its mobile phone service in Australia. Customers using the system are guided through the registration process by a speech recognition system.

In 2004, Vodafone Australia won a 'Global Best Practice Award' at the Conversations Convention in the US. It was the first time that this award has been won outside the US.