Skip to content

Introduction

das-Peak Streaming software is meant to cover registration and authentication of subjects in real-time scenarios in which the voice biometrics operation needs to happen simultaneously as the user speaks, in what is referred to as a case of agent-assisted registration or authentication. And preserving the strengths of the Veridas voice biometrics solution, das-Peak: authentication with only 3 seconds of speech audio, text, and language independent.

Today, the most widespread authentication method in Call Centers is called Knowledge-based Verification (KBV). A long (avg. 30 to 40 seconds) and insecure verification process consisting of asking the user for numerous predefined personal data questions to verify the end-user identity. More recently, other methods have appeared intending to reinforce security, such as the Phone-as-a-token methods (OTPs), resulting in a detriment in customer experience without bringing notable security improvements.

The scenario above describes how das-Peak Streaming comes to help to minimize the nuisance of the subject authentication process, increasing the overall system security and reducing the costs associated with this type of process. The agent could enable the subject identification in the background through the das-Peak Streaming voice Biometrics service. No time would be wasted with personal data questions or waiting for the user to dictate the OTP message received on his/her mobile phone. Instead, the user will just need to speak naturally with the agent, while he/she is being authenticated.

Right after picking up the call or through manual actuation by the agent, depending on the configuration chosen, das-Peak Streaming will immediately start recording the caller’s voice in the background, in order to accumulate just enough user’s voice to perform the voice biometric operation. Whenever this requirement has been satisfied, das-Peak Streaming will send the collected audio to the Veridas cloud das-Peak service, getting the registration/verification response outcome, and informing the agent of the same through the agent’s Call Center software user interface.

Figure 1. High-Level Architecture

For the cases in which the audio has already been recorded and stored as an audio file, such biometric operation doesn’t require to be done in real-time (while the audio is streamed), and therefore Veridas’ das-Peak service shall be used instead of das-Peak Streaming. Typical usage scenarios of this approach correspond to chatbots, IVRs, Whatsapp/Telegram apps, Virtual Assistants, etc.