Voice biometrics is an area of biometrics which focuses on identifying people by voice, or voice print as it is sometimes known. Biometrics in general involves measuring unique biological characteristics for the purpose of comparing unknown samples against known samples, usually with the goal of confirming someone's identity. This technology has attracted a great deal of attention in many regions of the world because it has potential for the security industry as well as other areas of human endeavor.
For a voice biometrics system, people must be enrolled in the system. People enroll by providing voice samples which are recorded and analyzed by the system. The more samples the system has, the more effective it will be, because having more samples allows the system to explore pitch, tone, speech patterns, stresses in speech, and so forth. A diversity of samples can reduce the risk of a false rejection by capturing more of the natural variation in a voice.
When someone who is enrolled speaks into a voice biometrics system for the purpose of identification, the system compares the voice against the voice prints it has on file. If it detects a match, it will indicate that the person has been recognized and identified. A display may bring up details about the person, and the system may also activate a door lock or similar device to allow someone to pass into a secure area.
When someone who is not enrolled approaches the system, the system should reject the individual after comparing the voice with the prints on file. One interesting thing about voice biometrics systems is that they are extremely hard to fool, with most having a higher false rejection rate than a false positive rate. Even if someone mimics the voice of someone else, the system can usually catch the differences. It can also identify recordings, which sound different from live voices.
Biometrics identification with a voice biometrics system is usually not impaired by colds, the sniffles, and so forth, because the underlying patterns of the voice are retained. Laryngitis or major throat surgery can sometimes alter the voice enough for the system to reject the person until she or he recovers, and in some cases, permanent voice changes may result in a need for reenrollment. These systems can also sometimes get confused if someone is angry, stressed out, or experiencing other intense emotions when she or he speaks for the system, as the strain can change the patterns of the voice enough for voice biometrics to be unsure about the identity of the speaker.
