Without using the words “Voice Biometrics” or “Voice Authentication” in its online documentation, Amazon Web Services (AWS) now offers its “Amazon Connect Voice ID” for contact centers. The service uses what AWS calls “machine learning” to provide text-independent caller authentication. It provides recommended procedures for “opt-in” enrollment that creates a digital voiceprint based on 30 seconds of natural speech from caller. When the enrolled individual calls back, Voice ID uses roughly 10 seconds of his or her spoken words to return a confidence score based on matching attributes of those utterances against a stored voiceprint.

Symnex Consulting’s Matt Smallman provides a detailed description and evaluation of the new service here. Matt provides a description of its key features; analyses for performance from the perspective of end users, contact center platform providers and go-to-market technology partners; and wraps up with a checklist of next steps that businesses can take to assess whether the service suits their needs.

Voice Biometrics “Lite” Isn’t New to Amazon

Amazon has used a form of speaker identification lodged in the cloud services that support its virtual assistant. In October 2017 , Amazon began supporting “voice profiles” for users of its popular Echo speakers to have Alexa distinguish between users. This solved a well known problem surrounding personalization. Alexa would play Dan’s playlists or recite items from Dan’s calendar as opposed to other family members. Its main rival, Google Assistant, introduced a similar service roughly 6 months prior.

Offering Voice ID as microservice rendered as an API in Amazon Connect’s control panel conforms to a classic marketing tactic by AWS. It will find organic demand among Connect’s user base and, by following the pay-per-use consumption model, provide a low-friction way for enterprises to pursue a low-friction authentication strategy. Most likely, security departments will not consider it strong enough to use as a single factor to initiate transactions. It is, however, well suited for detecting possible impostors as part of a multi-factor strategy.

Achieving “Default” Status: A Time-Tested Tactic for AWS

IVoice ID is rendered as a couple of rectangles that can be dropped into the call flow creator. Then AWS Connect’s default Contact Center Control Panel can prompt agents through both the enrollment and authentication processes by invoking a set of simple APIs. The barriers to get started are low or non-existent and pricing is simple, transaction-based and aggressive. A mere $0.025 for an enrollment as well as authentication. That’s disruption by default.

AWS Connect already offers Voice Biometric solutions from other vendors. Auraya Systems, for one, has been available through the AWS marketplace since mid-2020. Three other competing voice biometrics are also listing products there, including onelogin, Nomidio (which resells technology from Aculab), and Voicekey (UK-based neural network-based technology provider). Like many vendors in various Amazon marketplaces, these solution providers have learned that the cloud’s giant reseller will morph into a formidable competitor as soon as a market for the service is established.

The introduction of AWS Connect Voice ID marks a vote of confidence in Voice Biometrics. Amazon knows from monitoring sales, what its customers like about alternative vendors and their services. It has deemed that the time is right to offer its own flavor of Voice ID. This could mark a tacit validation of market for voice authentication.

Categories: Intelligent Authentication