Multitaper MFCC and PLP features for speaker verification using i-vectors

Abstract

In this paper we study the performance of the low-variance multi-taper Mel-frequency cepstral coefficient (MFCC) and perceptual linear prediction (PLP) features in a state-ofthe-art i-vector speaker verification system. The MFCC and PLP features are usually computed from a Hamming-windowed periodogram spectrum estimate. Such a singletapered spectrum estimate has large variance, which can be reduced by averaging spectral estimates obtained using a set of different tapers, leading to a so-called multitaper spectral estimate. The multi-taper spectrum estimation method has proven to be powerful especially when the spectrum of interest has a large dynamic range or varies rapidly. Multi-taper MFCC features were also recently studied in speaker verification with promising preliminary results. In this study our primary goal is to validate those findings using an up-to-date i-vector classifier on the latest NIST 2010 SRE data. In addition, we also propose to compute robust perceptual linear prediction (PLP) features using multitapers. Furthermore, we provide a detailed comparison between different taper weight selections in the Thomson multi-taper method in the context of speaker verification. Speaker verification results on the telephone (det5) and microphone speech (det1, det2, det3 and det4) of the latest NIST 2010 SRE corpus indicate that the multitaper methods outperform the conventional periodogram technique. Instead of simply averaging (using uniform weights) the individual spectral estimates in forming the multitaper estimate, weighted averaging (using non-uniform weights) improves performance.Compared to the MFCC and PLP baseline systems, the sine-weighted cepstrum estimator 2 (SWCE) based multitaper method provides average relative reductions of 12.3% and 7.5% in equal error rate, respectively. For the multi-peak multi-taper method, the corresponding reductions are 12.6% and 11.6%, respectively. Finally, the Thomson multitaper method provides error reductions of 9.5% and 5.0% in EER for MFCC and PLP features, respectively. We conclude that both the MFCC and PLP features computed via multitapers provide systematic improvements in recognition accuracy.

Multiple windowed spectral features for emotion recognition

“…More details about these three tapers can also be found in [10]. In this paper, we use the Thomson multitaper [7], the SWCE [14], and the Multipeak multitaper spectrum estimator [16] to compute the low variance MFCC and PLP features for an emotion recognition system.…”

“…For this study, we take the number of tapers, M, equal to 6 because it is found that this number optimizes the performance for speech recognition [12] and speaker verification problems, as reported in [10,11].…”

“…It has been shown in [10], in the context of speaker recognition, that the use of non-uniform weights instead of uniform weights (equal to 1/M) to obtain the final multitaper spectral estimate provides better recognition accuracy over the baseline. In the multitaper systems described in Table 1 (namely SWCE, multi-peak, Thomson_1 and Thomson_2), the final spectrum of the multitaper is estimated by averaging the M tapered subspectra using non-uniform weights.…”

“…The multitaper method has been widely used in geophysical applications and more recently, in speech enhancement applications [9] and in speaker and speech recognition [10][11][12] and it has been shown to outperform the windowed periodogram. Our main goal in this paper is to study whether the improvements achieved on speech and speaker verification tasks using multitapering, described in section 2, translate to the emotion recognition problem.…”

Lexical speaker identification in TV shows

“…State-of-the-art systems build on top of this basic framework by concatenating the GMM mean vectors to form speaker-specific supervectors, typically classified using Support Vector Machines (SVM) [7] or processed using factor analysis [10] or I-vector analysis [9,26,1].…”