For passive solutions, we offer our specialized text-independent voice biometric SDK called VTAthena.
VTAthena is based on state-of-the-art algorithms such as iVector/PLDA. VTAthena has been initially developed and engineered in close cooperation with reserach institute CRIM in Canada. CRIM pioneered many of today’s speaker recognition technologies such as Joint Factor Analysis and iVector paradigms.
VTAthena enables enrollment and authentication in a call-center environmnet with as little as 30 seconds of speech for enrollment and 8s of speech for authentication.
- Text-, language-, and channel independent
- State-of-the-art signal processing and statistical modeling algorithms
- Channel compensation for landline, mobile and data channel usage
- Voice quallity measurements
- Text-independence makes it applicable to arbitrary conversations
- Language-independence makes it applicable to multi-language call-centres
- Channel-independence allows usage of varying telephony channels
- 1:1 authentication and 1:N identification