[24]7 Adds Deep Neural Networks to IVR

[24]7 has launched a new version of its [24]7 Customer Engagement Platform that integrates Microsoft's Deep Neural Networks (DNN) technology into the interactive voice response (IVR) system. The addition is expected to yield speech recognition accuracy of at least 95 percent, an increase as high as 25 percent in some client application tests across several languages.

"We're excited to be the first company to apply deep neural networking to the IVR," says Patrick Nguyen, chief technology officer at [24]7.

That integration "is long overdue," he says, noting that deep neural networking has made its way into the consumer space with applications like Siri and Cortana, but hasn't been used in a business context until now.

Deep neural networking technology, as it applies to speech recognition, seeks to emulate the human brain when it comes to understanding and processing the spoken word.

 The [24]7 Customer Engagement platform with Microsoft DNN technology trains with analyses from more than 10 billion speech utterances from Bing search, Xbox, and Windows Phone and applies that training to enterprise self-service IVR interactions, which often include challenges like background noise, accents and dialects.

"With DNN, the speech recognizer is much more tolerant of outside noise, accents, dialects, etc.," Nguyen says. "We've been very surprised with the improvements in accuracy."

For the customer dialing into the IVR, deep neural networking will decrease the number of out-of-grammar errors, mismatches, and no-matches, Nguyen says, which should reduce the need to zero out or ask to speak to an agent. "As a consumer, you'll find that the IVR understands you more reliably," he says. "DNN will make the experience much better for the customer."

Adding deep neural networking to the IVR isn't expected to slow down the IVR, though it will require a little more computing power, according to Nguyen. "There will be no latency or difference in processing time," he says. "And from an industry standpoint, if you have up-to-date servers, you should be able to support it."

Avis Budget Group is the first customer to use [24]7's speech solutions with DNN technology across Europe. "Customer satisfaction and self-service are strategic priorities for Avis Budget Group, from reservations and confirmations to changes and extensions. It is critical that we provide the best, most efficient self-service experience possible," said Gerard Insall, chief information officer at Avis Budget Group, in a statement. "Our customers call from all types of noisy environments, and IVR systems have struggled in the past to make out the customer intent over and above the cacophony of airports and other busy public spaces. This new platform will help us to counter many of these issues and enable our customers to effortlessly complete their transactions in the IVR."

The integration of Microsoft’s deep neural networking technology into [24]7's IVR follows a close relationship the two companies have had for years. In early 2012, Microsoft and [24]7 announced a partnership that turned over Microsoft's enterprise IVR business to [24]7. As part of that deal, the two firms also agreed to a shared technology road map and a long-term intellectual property licensing agreement that provides broad coverage under Microsoft's patent portfolio for speech-related technologies. 

"The science behind Microsoft's Deep Neural Networks has dramatically improved speech recognition across a number of consumer and enterprise scenarios," said David Ku, corporate vice president of the Application and Services Group at Microsoft, in a statement. "We're pleased that [24]7 is able to leverage Microsoft speech technology to bring significant benefits to enterprise IVR."

And though [24]7 is the first to roll out the technology in an IVR, Nguyen expects deep neural networking to be the future of IVR.

"With the improvements in performance and accuracy we're seeing, it will increase people's comfort using speech self-service,” he says. "People will be much more willing to use IVRs when dealing with companies."

"Incorporating Microsoft's Deep Neural Networks into enterprise IVR is a gigantic stride forward for enterprises looking to deliver effective and reliable self-service for consumers," said PV Kannan, CEO and cofounder of [24]7, in a statement. "[24]7 has been an early innovator in applying advanced predictive modeling and data analytics to customer engagement. Adding DNN is a major milestone in that lineage. With the increasing proliferation of mobile devices and branded mobile apps, we know that more difficult journeys will go to self-service, so major companies need these next-generation solutions to meet customer expectations effectively."

SpeechTek Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues
Related Articles

[24]7 Launches Virtual Agent Technology

The virtual agent relies on technology gained during [24]7's acquisition of IntelliResponse.