KnuEdge Launches with Voice Authentication Software

After a decade of research and development using more than $100 million of private funding, former NASA head Dan Goldin has launched a neural technology innovation company called KnuEdge. The company's first output is a military-grade voice authentication product called KnuVerse.

Goldin, who headed up the U.S. space exploration agency from 1992 to 2001, founded Intellisis, which would eventually become today's KnuEdge, in 2005. KnuEdge is dedicated to developing technology and human-machine interactions, with a strong focus on machine learning and voice authentication and recognition.

"We started off partnering with the U.S. military for speaker identification," says Kate Dilligan, executive vice president of KnuEdge, who mentions that part of the KnuEdge's decade under the radar was due to the classified nature of itsstate-of-the-art technology. "While we've been developing the voice recognition technology for years, we finally reached a level of performance where we were comfortable with approaching a civilian market. It's taken time, of course, to get clearance for our proprietary technology. We had to apply for [International Traffic in Arms Regulations] approval, and that's taken quite some time. But the underlying technology for KnuVerse and the other KnuEdge systems has been field-tested."

KnuEdge's voice authentication and recognition can identify speakers in environments up to 80 decibels, and ocan operate in multiple capacities. Four initial deployments are available: Background, DNA, Active, and AudioPIN. Of the four, Active is the most versatile and traditional product, enabling users to authenticate individuals via utterance. Background lets users run constant voice recognition in enabled systems to identify individuals within conversational speech, with significant applications in archiving and searchable audio. DNA is useful for passive security, enabling users to tag and alert if certain voices are identified in audio streams such as incoming and outgoing calls, as well as video conversations. Finally, AudioPIN combines voice authentication with written and visual cues to create a secure biometric authentication that minimizes the threat of playback attacks.

"There is nothing unique that is required to implement KnuVerse voice authentication solutions, so getting to prototypes and production systems is very straight-forward. Within our single solution, we handle the complexity of audio captured from a range of devices, such as mobile devices, IoT devices, and laptops as well as standard phones," Dilligan states. "Companies need not plan for different packages. We offer simple and easy choices for integrating KnuVerse voice authentication into their existing systems with APIs delivered via the cloud or for on-premises integration."

The KnuVerse product line, including the KnuEdge voice authentication suite, was built to run in the cloud or to be hosted on a virtual private cloud. Voiceprints are encrypted at rest and best practices for secure data have been established both in-house and for dissemination to clients.

"We're aware that we're staking our reputations on the quality of how secure these products are," Dilligan says. "We're willing to have frank conversations with customers about their security needs. We're confident."

In addition to the robustness of the KnuVerse product suite, KnuEdge anticipates future-proofing its technology with the other two products available at launch: Knurld, its set of voice-authenticating RESTful APIs that help developers and clients implement the KnuVerse engine, and KNUPATH, a branch of KnuEdge that is producing processors optimized for scalability and machine learning. Currently, KnuEdge is marketing its voice authentication and its processors as separate products but anticipates bringing the two product lines together to increase capability as early as 2018.

SpeechTek Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues