Speaker Identification System Built on A Hybrid Model Through Different Feature Extraction Techniques-A Review


1Government Polytechnic College, Theni, India, 2FEAT, Annamalai University, India


This paper aims to provide a brief outline in to the area of speaker identification techniques. Speech is the most common way for human to communicate. The speech recognition permits system to interact and process the provided verbally by the user. Speech recognition can be defined as process of recognising the human voice to generate commands or word string. Speaker recognition can be divided in to dual techniques. Speaker identification and speaker verification. Speaker identification refers to the process of identifying human voice by means of artificial intelligence techniques. Speaker identification technologies are extensively useful in security and surveillance, voice authentication, electronic voice spying and identify verification. In the speaker identification process, extracting noticeable features from speaker utterances is an important task to accurately identify speakers. Deep leaning algorithms have been mostly used to further enhance the capabilities of computers excellently.


MFCC, Spectrogram, GMM, CNN, DNN