Automated Tuning Project

ArmorVox uses machine learning algorithms to tune the background voice print models to match its production environment. This ensures that optimal results are achieved for all types of speakers on all types of channels, allowing a more accurate and secure result.

Speaker Specific Threshold and Speaker Specific Background Models (Patented)

ArmorVox uses machine learning algorithms to allow clients to set specific thresholds per user or use case. With this option, clients can set security settings to set a baseline security or audit a singe user’s security levels if deemed not strong enough. This ensures a consistent level of security for every user every time.

Active Learning (Patented)

ArmorVox™ uses machine learning algorithms to continually learn from each user interactions. By enabling active learning, ArmorVox™ is able to improve voice prints for each individual user, allowing a more secure and accurate verification. This also helps improve the performance of ArmorVox™ over time across all channel implementations.

Impostor Mapping (Patented) and Cross-Matching

ArmorVox™ uses machine learning algorithms to wash every new enrollees against the current database to identity potential duplicates in the system or even detect impostors and fraudsters. On a single server, ArmorVox™ is able to cross-match users in real-time with a cross-match speed of over 25 million voice prints per hour.

Fused Passive/Active Modes

ArmorVox™ allows clients to use either text-dependent, text-independent, text-promoted or digit-independent voice prints for customer enrollments. This allows clients to implement a secure voice identification and verification solution that is tailored to their use cases. Most importantly, this gives the clients and users the option to enroll and verify with whatever method or process is most comfortable and easiest for them.

Synthetic Voice Detectors

ArmorVox™ uses machine learning algorithms to detect and repel synthetic voice systems. This feature allows clients to detect and reject generated voices or fake voices and even recorded voices to ensure a secure and accurate capability.

Liveness Detection

ArmorVox™ can deflect pre-recorded responses by issuing randomly generated challenge responses. This means that clients can set these challenges to mitigate impostor or fraudulent activity via pre-recorded messages as verification processes can be randomized.

Omni-Channel

Implement ArmorVox™ in either telephony or digital channels. ArmorVox™ comes as an API which allows clients to install our solution easily into any devices and platforms. Let users enroll or verify using their voice during conversations in call centers or while using a web browser, mobile application or chat bots. Improving customer experience and security is as easy as that!