Text independent speaker recognition using the Mel frequency cepstral coefficients and a neural network classifier

Principal Component Analysis Based Speaker Verification

“…A functional way to think about this filter bank is to view each filter as a histogram bin (where bins have overlap) within the frequency domain. Figure belo w gives an examp le of a mel-spaced frequency bank [4]. We have used 20 filters in my work.…”

MFCC and Prosodic Feature Extraction Techniques: A Comparative Study

“…The MFCC is a type of wavelet in which frequency scales are placed on a linear scale for frequencies less than 1 kHz and on a log scale for frequencies above 1 kHz. MFCC is capable to capturing the important characteristic of audio signals [1] [5] [7].The complex cepstral coefficients are called the MFCC. The MFCC contain both time and frequency information of the signal and this makes them more useful for feature extraction.…”

“…Both spectrumbased speech features, related to the shape of the vocal tract, and prosodic features, related to the excitation of the vocal tract and the speaking style of a person. In Speaker Recognition the extraction and selection of the best parametric representation of acoustic signals is an important task in the design of any speech recognition system; it drastically affects the recognition performance [7]. There are many studies have been done on MFCC and Prosodic which tell us that the MFCC use A small set of standard features while Prosodic uses long term features.…”

“…Therefore the number of filters can be ads or subtracts to provide the sampling frequency of experiments. The performance of the Mel-Frequency Cepstrum Coefficients (MFCC) may be affected by the number of filters and type of window [2] [7] i.e. accuracy rate fluctuate to increase the number of MFCC coefficient and also to decrease the MFCC coefficient.…”

“…In addition too few or too many filters do not result in better accuracy. Also is described in [7] that efficiency is maximum while using hanning window.…”