Speech Analytics: Showing Great Promise Thanks to AI and Machine Learning

Article Featured Image

Speech analytics has been commercially available for about 15 years, making it a mature technology, but it is only now starting to move beyond the pilot and early-adoption stages, according to industry experts.

To demonstrate just how slow progress has been thus far, nine in 10 respondents in a 2017 Opus Research survey said they had incorporated speech analytics into their customer care framework, but just a little more than a third said they had more than a year of experience in using the technology.

And though the technology is only now starting to build a following, growth is not expected to be as robust as it is for other speech technologies. DMG Consulting, for example, predicted the contact center speech analytics market would grow by 12 percent in 2017. In 2018, it predicts growth at 11 percent, and puts growth at 10 percent per year in 2019, 2020, and 2021.

But that isn’t necessarily a harbinger of bad times for the technology. “The future of speech analytics is positive. Adoption remains strong, and there is still significant unpenetrated market for stand-alone solutions,” DMG concluded in a recent report. “An even greater opportunity exists for vendors to deliver solutions that are cloud-based and address both speech and text interactions.”

Donna Fluss, DMG’s president, divides the current speech analytics landscape into two types of vendors: a large group of competitors who sell basic applications that focus on identifying keywords and phrases, and a smaller set of feature-rich solution providers who offer extensive business intelligence platforms that provide enterprise-level data. The greater market potential likely exists for the latter, though solutions today are still mostly being used by contact centers to convert unstructured phone conversations into structured data that can then be used to make inferences about larger company trends and challenges.

The addition of artificial intelligence and machine learning into current speech analytics solutions will help in that regard and also bolster technology adoption, according to Fluss.

“The next step for speech analytics is to enter the world of data science. In the future, artificial intelligence [AI] will be used to enhance the ability of speech analytics to identify issues and recommend ways to address them,” she points out. “Machine learning is starting to be used to enable these solutions to quantify the impact of new trends and issues, with minimal human intervention. The opportunities and benefits are substantial for vendors and end users. Companies that take speech analytics seriously and invest in the resources and best practices to build an effective program are realizing significant benefits.”

Dimitris Vassos, CEO of speech company Omilia, agrees. 

“[In 2018] a new breed of speech analytics will rise. This new breed of speech analytics understands concepts and context instead of just keywords, monitors the dialogue of both virtual and human agents, and takes real-time actions to recommend, prevent, and augment the customer experience,” he predicts.

Many of the leading vendors of speech analytics technology are already moving in that direction. In October, Verint Systems, for example, introduced automated speech analytics powered by machine learning. Analyzing all contacts across individual customer journeys or identifying contacts from multiple callers with the same language can help analysts understand the root cause of repeat contacts, according to Verint. That information can then trigger swift corrective action to expedite resolution.

Calabrio is also experimenting with machine learning as a way to help companies predict human behavior, such as whether a customer threatening to drop a service is bluffing or is actually planning to churn. Calabrio’s solution combines speech and text analytics with human behavior to help guide agent responses, according to Matt Matsui, senior vice president of product and strategy at Calabrio.

Another industry-wide change that is helping to steer speech analytics toward greater adoption is a renewed focus from end users.

More Than Just Cool

“In 2018, companies are [shifting attention] away from the cool elements of the technology to a focus on the actual solution,” says Jeff Gallino, founder and chief technology officer at CallMiner. “The growth of the use of personal digital assistants has driven the need for a better quality of speech recognition.”

This focus has also shifted from reactive to proactive, Gallino adds. “You can do predictions in-call, in real time, rather than just analysis for the next call.”

With modern speech analytics, contact center agents can get real-time notifications when callers use certain keywords or phrases that might make them good prospects for a particular product or service or when certain offers might help mitigate their complaints. 

Such knowledge is critical today, Gallino says, because even with the growth of digital communications and online chat, phone calls still represent a critical element of customer care.

“Everyone is predicting the death of phone calls,” Gallino explains. “They predict the decline of phone calls every year, but that hasn’t happened, even with the rise of new channels. Speech as a data source is not going to go away; it’s only going to grow.”

Another area of great growth potential for speech analytics is workforce optimization (WFO), according to Fluss. WFO vendors are now starting to support more nontraditional channels and adding new functionality, with an emphasis on intelligent automation and robotics to improve front- and back-office business functions. 

“It’s going to be an interesting three to five years for the WFO sector as it undergoes a major shake-out,” Fluss says. “Vendors who can transition to new-gen platforms designed to support the digital economy will succeed, while others who cannot respond to market changes will quietly try to find an exit. This transition opens the door for emerging solutions that take a digital, omnichannel, analytics- and AI-oriented approach to addressing the WFO challenge.”

Other companies are using speech analytics as a critical component of their efforts to fight fraud in the contact center and to guarantee that agents are adhering to prepared scripts and complying with all applicable rules and company policies.

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

The Power of Voice in Health Care

The Voice Health Summit explores all the ways speech technology and AI are transforming healthcare.