“The Death of Voice” as a channel for customer care is not just greatly exaggerated, it is downright incorrect and misleading. large contact centers routinely generate billions of minutes of recorded speech every month. Customer care professionals and contact center managers well know that the content of conversations between agents, representatives and other professionals is an asset that is of great use for training, compliance and even sentiment detection. A partnership between analytics specialist CallMiner and SpeechPro tackles one of the major known deficiencies in a number of contact center architectures – distinguishing between a caller and a company representative.

The term “speaker separation” may be new to many readers. It is something that the human mind does almost automatically, based on the direction a voice comes from or other unique attributes of a speaker. With so much material to analyze, executives have put some of the most sophisticated speech recognition and analytics resources to task to achieve their business goals.

Computers can do at scale what we humans do on a personal basis but most “legacy” call recording systems fail to distinguish between the caller and the company representative. This makes it much more difficult to the types of speaker-based categorization that support compliance and training efforts for agents or sentiment detection or intent recognition for callers. SpeechPro’s VoiceGrid product suite includes proprietary technology to distinguish between individual speakers in a recorded conversation.

Voice biometrics already plays many roles in contact centers around the world. Simple, speedy voice-based authentication is the most conspicuous, with over 70 million people using voice passwords to identify themselves. Fraud loss prevention is another area where voice biometrics has a major economic effect. “Speaker separation” sounds arcane and a bit sterile but, as noted above, recognizing a callers intent – even if you don’t associate it with a specific caller’s identity – is a big part of Intelligent Assistance. You can’t provide personalized service without distinguishing the caller’s voice from that of an agent or representative.

Categories: Intelligent Authentication, Articles

Tags: speech analytics, Speaker recognition, Intelligent Assistance, conatact center, speaker separation