Speaker identification system based on a web interface

We present a new web-based application designed for human computer interface that currently supports speaker identification module. It is based on Java EE and Spring Framework and is designed to be invoked by users through their Internet browsers. Due to a flexible design various feature extraction methods, signal processing and classification algorithms can be easily implemented and used in different combinations. Apart of a brief introduction of the system architecture the main focus is on a set of experiments optimizing the overall functionality, i.e.: aiming to improve the accuracy, robustness and time performance. However, some theoretical bases of speaker identification problem are presented as well.