A comparison of four techniques for automatic speaker recognition

Abstract

Four automatic speaker recognition techniques were investigated with a conunon speech data base to determine their effectiveness in a text independent mode. These four techniques used the correlation of short and long term spectral averages, cepstral measurements of long term spectral averages, orthogonal linear prediction of the speech waveform, and long term average LR reflection coefficients carbined with pitch and overall power. The results of this study indicate that LC derived parameters perform better than do those derived from cepstral and spectral data. Recognition accuracies of 95% and 93% were obtained for LFC based techniques with 13 seconds of unknown speech. The corresponding recognition accuracies for the cepstral and spectral based systems were 79% and 54% respectively. INTJDTION The purpose of this study was to test and caipare four techniques of automatic speaker recognition on a conmon data base. Performance was caipared in a text independent environment on free conversational speech. This paper details the methods used and the results obtained during the study. The four techniques ccnipared are: 1. The correlation of short and long term spectral averages as investigated by S. Pruzansky and M.V. Mathews [11. 2. Cepstral measurements of long term spectral averages as investigated by S. Furui, F. Itakura, and S. Saito [2]. 3. Orthoqonal linear prediction of the speech waveform as investigated by M.R. Sambur [3]. 4. Long term average LPC reflection coefficients, pitch and overall gain of the speech waveform as investigated by

Methods of Combining Multiple Classifiers with Different Features and Their Applications to Text-Independent Speaker Identification

“…Although Wolf outlined a set of desirable attributes on the chosen features for speaker recognition [64] more than 20 years ago, unfortunately, it is highly unlikely to find any set of features which simultaneously has all those attributes in practice [4,18,21,22,24,28,44,49,52,56]. As a result, several features have already been investigated [4,21,22,28,34,63]. The main outcome of the many feature selection studies was that features which represent pitch and the speech spectrum were the most effective for speaker identification.…”

“…In particular, the problem becomes quite serious when the techniques of neural computing with timedelay [6,8,9,13,62] are used. On the other hand, several kinds of classifiers have been also applied in speaker identification [9,18,24,28,49,63]. These classifiers include distance classifiers [3,4,25,33,42], neural network classifiers [6,7,8,11,12,13,14,19,32,46,47,54] and classifiers based upon parametric or non-parametric density estimation [28,29,52,57,59].…”

International Journal of Electronics and Telecommunications

“…Features extraction various methods, used by many authors, include the basic techniques, based on spectral average values [4], pitch of tone, coding with linear prediction (LPC) [5], cepstral indices on the basis of the linear prediction (LPCC) [6], mel-cepstral coefficients (MFCC) [6] , etc. [7] Speech recognition technology is widespread in different business areas: -solutions "Smart house": voice interface of managing the system «Smart house»; -household appliances and robots: electronic robots voice interface; voice control over household appliances, etc.…”