How Is the Voice Captured?

The voice can be captured either through the use of dedicated resources or existing infrastructure. An example of a dedicated resource is a stick microphone attached to a computer. Existing infrastructure can take the form of a telephone.

The use of either type of capture device is influenced by two qualities. The first is the physical quality of the device, and the second is the environment in which the sample is being taken.

For a microphone connected to a computer, the higher the quality, the better. As the quality of the microphone improves , its dynamic range increases , as does the quality of the components being used. It may also include some noise-reducing technologies to remove background ambient noise. If it is a multidirectional microphone, it is more likely to pick up background noise. If it is unidirectional, then it will normally take only sound waves originating from a particular spatial region.

Most phone sets are normally just sufficient for voice conversations. They can have poor microphones in the mouthpiece, or be of older, less advanced technology. In addition, the housing of the handset may introduce noise and distortion. The set itself may not even be digital. If it is analog, then the raw signal propagated down the line will be susceptible to line noise and distortion. The use of a digital set would convert analog signals to digital signals before transmitting. In this way, the quality of the voice signal can be maintained , as its digital representation will be used as input to the recognition algorithm. Like a microphone, ambient noise can affect the quality of the capture. Again, the use of a unidirectional microphone can help with this. It is generally not advisable to do voice biometrics with a speakerphone!

With the prevalence of wireless phone technology, these could also be used as capture devices for voice biometrics. Since the signal needs to travel through the open air and then to terrestrial lines, it is very susceptible to interference. With the current analog cellular technology, the quality of the voice can approach that of tin cans and string. The quality of the sound and signal is significantly improved with the newer digital cellular technology, which may make it a viable source for voice biometrics.