KSV announces voice biometrics deployments, Speechmatics signs two new speech recognition partners
Voice biometrics from Kaizen Secure Voiz (KSV) are being used by a national security firm with hundreds of employees, more than half of whom are mobile guards in the field, to track the time and attendance of personnel scattered among numerous organizations across the country.
Another customer of KSV’s IVR-based time and attendance system is a hospital with hundreds of doctors and healthcare workers making up its workforce, and three-quarter of those working in the field.
Mobile employees dial a toll-free number, utter a language and text-independent passphrase, and are authenticated in less than 10 seconds, the company says. The voiceprint is passively set, and caller identity can be verified inexpensively, efficiently and securely, according to the announcement.
“An enterprise can very easily adopt Kaizen Secure Voiz voice biometric product lineup when a customer reaches out to their contact center for any service or you can ask for explicit consent from customers and proactively enroll them,” states Ashok Krish, President of Kaizen Secure Voiz. “It’s user-friendly, safer, cost-effective and with user education can help one overcome any challenges faced.”
KSV’s patented technology also supports contactless processes and pandemic-related safety precautions.
Speechmatics speech recognition integrated by Veritone and Prosodica
Speech recognition technology from Speechmatics has been integrated with Veritone’s aiWARE operating system to provide a secure transcription capability, and by Vail Systems subsidiary Prosodica to ensure security and compliance throughout its system.
Speechmatics technology enables Veritone to help customers with strict security needs convert unstructured voice data into actionable insights. That includes users of Veritone’s FedRAMP-approved environment for the U.S. government.
The aiWARE platform helps enterprises and other organizations automate their processes and workflows by providing hundreds of cognitive services in one place, the companies say. Insights from voice data can empower organizations to make data-driven decisions, improve efficiency, find new opportunities and manage compliance risks.
“Only in the last few years has speech-to-text technology become accurate enough for businesses to truly utilize, incorporate it as part of their business workflows and deliver material benefits,” says Speechmatics VP Jeff Palmer. “Speechmatics is the most accurate, any-context speech recognition engine on the market, whilst also offering a broad range of languages and the secure, on-premises deployments that are essential for our customer’s needs. Veritone offers the most comprehensive operating system for AI in the market so naturally, we are thrilled to be partnering with them. Veritone’s FedRAMP Authority To Operate (ATO) means we are able to operate fully across both Veritone’s US federal government and worldwide commercial customer base securely and accurately as a cognitive application.”
Prosodica provides voice and audio analysis to enhance customer care and experiences. Integrating with Speechmatics’ any-context speech recognition technology enables the generation of next-level insights from recordings, according to a separate announcement.
“The voice and speech analytics capabilities in the Prosodica platform are enabling our customer to transform their operations and customer experiences,” Prosodica President and CEO Mariano Tan comments. “Speech recognition technology is essential for the functionality of our applications, so we teamed up with Speechmatics to ensure we deliver the best speech technology on the market.”
Limits of voice analysis considered
The limitations of voice biometrics and analysis, and whether attempting to identify risk from a person’s voice goes beyond them, is examined by the ABC’s Radio National.
In light of the use of voice biometrics by the Australian Tax Office (ATO), the ABC considers the history of voice analysis, from early efforts related to eugenics through experiments by Bell Labs in the forties. Australia began using language analysis for the determination of origin for asylum applicants without documentation in the early 2000s, which University of Melbourne Applied Linguistics Professor Tim McNamara says was problematic in practice.
Clearspeed claims to be able to detect risk in people’s voices, in the form of “cognitive effort” when answering an automated questionnaire. The idea is to be a tool for quickly approving most applicants, such as for a job interview, and send a small number for manual review.
ABC also notes Descript’s new Overdub software, which the company says clones a subject’s voice from 10 seconds of speech. The results are not exactly deepfakes, ABC reports, but should certainly be taken into account by anyone selling voice analysis technology to prevent fraud.