Voice versus Speech
Voice Biometrics compared to Speech Recognition
People often confuse the ArmorVox™ Speaker Identity System with common speech recognition products. After all, from a user’s point of view, the two technologies appear similar. However, they do very different things.
Speaker Identification is a Security Solution
Voice authentication systems, such as ArmorVox, integrate into an organization’s security and identity management infrastructure. Speech recognition technologies usually integrate into the communications channel. This distinction, whilst subtle, had a profound impact on the way ArmorVox was developed and architected.
Speech Recognition is for Dictation and Command
Speech recognizers are commonly used in dictation (speech to text transcription) and voice command applications (such as car or smart phones). Many voice biometric products were derived from speech recognition technologies. They are often delivered, configured and supported as if they were speech recognition products. This introduces potential security risks, which can adversely affect the system, perhaps denying access to true speakers or allowing impostors unauthorized entry.
The distinction between “who is speaking” and “what is being said” is fundamental.
Speech recognition is a core human-computer communications technology, attempting to understand “What you are saying, irrespective of who you are.” ArmorVox is a voice biometrics security technology, working out “Who you are – sometimes irrespective of what you are saying.”
They are, if you like, flip sides of the “voice technology coin.”
Auraya Systems designed ArmorVox to be language independent from its inception. ArmorVox can operate in regional markets or in multi-lingual global markets out-of-the-box, unlike other speech recognition technologies that need to be tuned and optimized for specific languages, accents and dialects.
ArmorVox is a Security Technology
ArmorVox™ Speaker Identity System was built from the ground up as a security technology with these functions:
- Security calibration for each identity registered on the system.
- Voice quality checking to eliminate vulnerabilities introduced by poor voice samples.
- Automatic tools to set thresholds and business rules based on the security performance for each identity registered on ArmorVox.
- Tuners to maintain optimum security performance for each speaker registered on ArmorVox.
For organizations who are implementing enhanced security in their customer call centers, help desk, monitoring and other biometric security applications, in the cloud or in your own premises, the ArmorVox™ Speaker Identity System is a natural choice.