Product overview
Why Containers?
The following points describe why LumenVox has adopted container technology:
We listen to our customers’ feedback
LumenVox works closely with our customers and partners and listens to their requirements and feedback on our technology stack. There has been a growing need to provide our full product offering through a containerized architecture.
We adopt new technology
We aim to stay at the forefront of technology that customers use and need, we aim to provide enablement through technology. We continue to provide the best, most highly performant, reliable products for customers and partners.
We empower customers to implement a myriad of use cases & minimize deployment challenges
Containers allow customers to manage a large array of speech and voice biometrics use cases and minimize the challenges with integrating them. Traditionally, integration, scaling and update testing are the largest challenges. Our new architecture allows for significantly simplified deployments.
We provide high performance solutions
The new architecture uses a highly horizontally scalable approach that provides an easy deployment and upgrade solution. It also provides self-healing and auto-scaling, ensuring that our services are always available.
We meet customer requirements for cost-effective infrastructure options
Customers can use existing machines and connectivity or migrate to more cloud-native technologies. The architecture also makes use of free software to support the solution e.g., PostgreSQL and Mongo as opposed to the costly traditional Microsoft SQL or Oracle database solutions, thus reducing the overall running costs.
Active Voice Biometrics Channels
The LumenVox Active Voice Biometrics product is available for use through any self-service channel, the customer simply needs to obtain the required audio samples and communicate with the APIs available. The channels that can be used include:
- IVR (inbound or outbound) utilizing mobile or landlines
- Smartphone application
- Web application
- Personal virtual assistant
- Desktop application
- Video calls (if the audio can be stripped out)
- WhatsApp and other messaging platforms
- Multimodal chatbots and Conversational AI apps
Product Use Cases
Voice biometrics can be used for the following use cases:
- Proof of life
- Password reset
- Caller authentication into a contact center
- Self-service IVR transactions
- Login to websites or smartphone applications
- Credit card suspicious transaction approval
- Transaction approval
- Step-up approval of high value transactions
- Portal login: the consumer enters an identifier (account number/Identity number) and their registered cellphone receives a callback for verification
- Digital collections: the consumer receives an SMS and clicks on a link through to a secure URL where they can authenticate themselves using their voice
- E-filing (tax office): confirm/validate biometrically that a person is either accessing their tax profile and/or authenticating their submission
- Online sales: e.g., insurance (taking out a policy). The consumer is required to enroll their voice to accept terms and conditions
- Registration using multi modal biometrics – where voice is added to face, video, finger-print, and other biometric means of authentication
- Electronic signature – the consumer uses their voice to digitally sign a document
- Proof of presence – use voice with geo-location to verify proof of presence e.g., home incarceration
Voice Biometrics Print Enrollment Process
Each consumer is required to repeat a specific passphrase e.g. “My voice is my password” a minimum of three times (3 separate utterances) to enroll their voice against a claimed unique identity e.g., their mobile number, account number or national identity number. The three utterances are used to create and store a voice model against the claimed identity.
Identity Verification Process
Should the consumer want to authenticate themselves, they are required to repeat the specific passphrase used at enrollment e.g. “My voice is my password”. This utterance is then matched against the voice model stored against the speaker’s claimed identity. Should the utterance score be above the pre-determined threshold then it is deemed a match. If the utterance scores below the pre-determined threshold, it is deemed a mismatch and the client should either reject the transaction or fall back to a secondary authentication process. Audio will be rejected if certain audio quality checks fail e.g., audio too soft, too short, or too noisy. The passphrase is also validated using LumenVox’s ASR engine to determine if the speaker repeated the correct phrase.