I. Introduction
Automatic speaker recognition is a technique used to automatically recognise the identity of a speaker from a recording of their voice. Speaker recognition is an important topic in signal processing and has a variety of applications, especially in security systems [1]. Voice controlled systems and devices rely heavily on speaker recognition. Speaker recognition consists of two fundamental tasks, namely speaker verification and speaker identification. Speaker verification is the task of determining whether an unknown voice is from a particular enrolled speaker. The speaker in this case provides a voice sample with a claim to be one of the enrolled speakers and the system either rejects or accepts the claimed identity. Speaker identification is the task of associating an unknown voice with one from a set of enrolled speakers. The speaker provides a voice sample and the system determines to which of the known set of speakers the voice sample belongs. Speaker identification systems can be classified based on the system range of operation, the classification can either be closed-set or open-set [2]. In a closed-set identification, the speakers are all enrolled into the speaker database and the speaker with the closest match to the test signal is chosen to be the test speaker. In the case of open-set identification, speakers need not always be enrolled into the database, thus the system needs to perform an additional task of rejection in case the speaker is someone from outside the speaker database [2].