Nuance Boosts Accuracy Of Speech-Recognition CallsNuance Boosts Accuracy Of Speech-Recognition Calls

The latest version of Nuance's speech-recognition software adds features aimed at improving the accuracy of automated speech-recognition calls.

Nuance 8.0 features auto tuning, which, over time, improves the system's accuracy in recognizing similar dialects or accents within a certain group of callers. In addition to becoming familiar with common accents, the system can also recognize callers who access the system via cell phones or from noisy environments, eventually making adjustments to filter background noises, says Marcello Typrin, Nuance senior product marketing manager.

The enhanced system also includes a personalization feature that detects a caller's gender, whether the caller is using a cell phone or landline, and measures how softly a caller is speaking in comparison to background noises. "When the system detects a soft-spoken individual, it prompts the system to say 'Can you speak louder,'" Typrin says. Gender detection also lets companies that use the system in their call center directly market products or services to customers. For example, a retailer could program its automated speech system to market upcoming sales on women or men's fashions, Typrin says. Nuance 8.0 is available now, priced at $200 per port.

Improving the accuracy of a call using an automated speech system can result in significant cost savings for businesses, says Dan Hawkins, voice business program manager with consulting company Datamonitor. Companies can use automated services to cut costs because they don't have to pay wages to live agents, Hawkins says. Improvements in automated services could also lead to higher sales. "Customers find it annoying when speech recognition doesn't work and they can't get to the information they need," Hawkins says. "The more you can minimize the times the system doesn't understand you, the better the experience for customers."