Voice Biometrics - Technical Outline

Like the other biometric markers, the result of a biometric measurement of the voice is totally dependant on

 1. input,

2. accurate mathematical algorithms, and

3. computing power. 

 Input refers to the biometric sample, such as a voiceprint, taken and stored in a database. 

 Input quality, the most important factor, is greatly affected by the type of input device (professional microphone v. cell phone, for example) and environment (noisy street vs. quiet office). CellMax Systems technology automatically measures voice sample quality, then corrects and cleans, whenever possible, to produce the clearest possible data.

 Algorithms are a set of precise steps that describe a limited procedure or task. Algorithms in biometric systems are used to find out whether a sample matches the stored input. The more precise the algorithm, the more accurate the matching process. 

 Levels of accuracy are measured in terms of False Acceptance Rate (FAR)/ False Rejection Rate (FRR).

 • J. Markowitz Consultants defines false acceptance as “when a speaker-verification application allows an impostor to get in.” False rejection is “when a verification system rejects a valid user.” 

 • FAR refers to the probability that a biometric system will incorrectly identify a valid user, or will fail to reject an impostor. FRR refers to the probability that a biometric system will fail to identify a true enrollee.

 • Real-time algorithms refer to algorithms that process information and return results so rapidly that the interaction appears to be instantaneous.

 Computing refers designing system to process voice biometrics data efficiently so that individuals are quickly identified and verified, or rejected.

 The voice biometrics solution developed by CellMax Systems takes as its foundation these 15 parameters that create a personal voiceprint. It then makes calculations using a proven, real-time mathematical algorithm with an unprecedented rate of accuracy: one-to-one verification that can extend beyond 99.8% and one-to-many identification of up to 98%.

Voice Biometrics - Technology

 CellMax Systems utilizes a voice verification algorithm to provide an improved method and system for registering and authenticating secure, voice-based, e-commerce transactions over telecommunications networks. 

 The technology provides a method and system for voice registration involving three major steps: 

• fractal analysis

• spectrographic analysis

• determination of Lyapunov exponents (see Glossary)

The method performs fractal analysis, where raw data is investigated and each sample gives a set of non-dimensional numbers that characterize a speaker's voice uniquely. 

 The method also produces a vector consisting of the aforementioned 15 physical voice parameters that form the key index for the verification. 

 The system performs a spectrographic analysis, investigating the raw data to generate a uniquely identifiable pattern. 

 The system includes a voice registration unit for providing unique initial identification by finding the speaker/user's voice parameters in a voice registration sample and storing it in a database. 

 The system also includes a voice-authenticating unit for verifying one of a plurality of users. The voice-authenticating unit includes a recognition unit for providing a voice authentication sample that operates with the database. The voice-authenticating unit also includes a decision unit that operates with the recognition unit and the database, to decide whether the user is the same as the person of the same identity registered with the system. In this way, the user’s identity is verified absolutely.