After seeing many kinds of technologies that involve corneal recognition in some high tech movies and finger print recognition in latest Apple devices. However, you probably don’t know speaker recognition, which is also used to recognize and confirm the identification of a person.
Speaker recognition can basically be divided into some steps. First, the model training step, in which a certain piece of speech spoken by a person will be recorded, after signal acquisition, manipulation and modeling, a unique model will be generated for the certain person. Then the model can be stored in a database for further use. In next step that can be divided into two tasks, speaker verification or speaker identification. Speaker verification is that to determine from a voice sample if a person is whom he or she claims. Speaker identification is to determine which one of a group of known vices best matches the input voice sample.
In our master period, we will doing a thesis about speaker identification using Gaussian mixture speaker models. We will not only implement this algorithm in C, but also analysis the results of record in different noisy environments, and try to give some solution to increase the identification rate and robust the system.
In the period where full of smart devices, we seem getting bored in bigger screen and faster CPUs. In our opinion that the “smarter” devices is the next goal of development. We will using our voices, gestures and maybe our brains to control our devices instead of using fingers. All in all, speaker recognition must play a very important role at that time.
 Douglas A. Reynolds, Member, IEEE, and Richard C. Rose, Member, IEEE, “Robust Text – Independent Speaker Identification Using Gaussian Mixture Speaker Models” in IEEE Transaction on speech and audio processing, VOL. 3, NO. 1, January 1995.