Performance Of Speaker Recognition System Research Articles

Utterance partitioning for speaker recognition: an experimental review and analysis with new findings under GMM-SVM framework

The performance of speaker recognition system is highly dependent on the duration of speech used in enrollment and test. This work presents a detailed experimental review and analysis of the GMM-SVM based speaker recognition system in presence of duration variability. This article also reports a comparison of the performance of GMM-SVM classifier with its precursor technique Gaussian mixture model- universal background model (GMM-UBM) classifier in presence of duration variability. The goal of this research work is not to propose a new algorithm for improving speaker recognition performance in presence of duration variability. However, the main focus of this work is on utterance partitioning (UP), a commonly used strategy to compensate the duration variability issue. We have analysed in detailed the impact of training utterance partitioning in speaker recognition performance under GMM-SVM framework. We further investigate the reason why the utterance partitioning is important for boosting speaker recognition performance. We have also shown in which case the utterance partitioning could be useful and where not. Our study has revealed that utterance partitioning does not reduce the data imbalance problem of the GMM-SVM classifier as claimed in earlier study. Apart from these, we also discuss issues related to the impact of parameters such as number of Gaussians, supervector length, amount of splitting required for obtaining better performance in short and long duration test conditions from speech duration perspective. We have performed the experiments with telephone speech from POLYCOST corpus consisting of 130 speakers.

Performance of speaker recognition system using shifted mfcc, delta spectral cepstral coefficient (DSCC) and Fuzzy techniques

Speech and speaker recognition systems are biometric inspired systems which are having scope in various online and offline applications. In case of biometric we ponder the variability of speech signal due to the presence of noise which greatly degrades the efficiency of Automatic Speaker Recognition (ASR) in real-world environmental circumstances. Real world speech signal is degraded by different types of noise signals like background noise, interference noise and crosstalk noise. In this paper, we have used Delta Spectrum Cepstrum Coefficient (DSCC) and Shifted MFCC with fuzzy modeling techniques to rectify the deed of ASR even in a noisy surrounding with the help of upgraded speech information which is present at high frequency in the spectral domain. The combination of fuzzy modeling and DSCC creates a firm cumulative algorithm which has reasonably high robustness to noise. Experimental results show that accuracy has enhanced by 10-20% even at 5-8dB SNR in the presence of background noise or turbulent environmental condition or in the presence of white noise.Thus proposed model has improved maturity level in comparison to obsolete methods.

Cost-sensitive learning for emotion robust speaker recognition.

In the field of information security, voice is one of the most important parts in biometrics. Especially, with the development of voice communication through the Internet or telephone system, huge voice data resources are accessed. In speaker recognition, voiceprint can be applied as the unique password for the user to prove his/her identity. However, speech with various emotions can cause an unacceptably high error rate and aggravate the performance of speaker recognition system. This paper deals with this problem by introducing a cost-sensitive learning technology to reweight the probability of test affective utterances in the pitch envelop level, which can enhance the robustness in emotion-dependent speaker recognition effectively. Based on that technology, a new architecture of recognition system as well as its components is proposed in this paper. The experiment conducted on the Mandarin Affective Speech Corpus shows that an improvement of 8% identification rate over the traditional speaker recognition is achieved.

Speaker recognition method based on Mel frequency cepstrum coefficient and inverted Mel frequency cepstrum coefficient

To improve the performance of speaker recognition system,a new method of feature extraction was proposed based on Mel Frequency Cepstrum Coefficient(MFCC) and Inverted MFCC(IMFCC).This method constructed a mixed feature by combining MFCC with IMFCC using Fisher criterion.The experimental results show that the mixed feature proposed in this paper has better recognition performance compared with MFCC not only in the pure voice database but also in the noisy environments.

基于自适应高斯混合模型与静动态听觉特征融合的说话人识别

By optimizing the feature vectors and Gaussian Mixture Models(GMMs),a hybrid compensation method in model and feature domains is proposed.With the method,the speaker recognition features effected by the noise and the declined performance of GMM with reducing length of the training data under different unexpected noise environments are improved.By emulating human auditory,Gammatone Filter Cepstral Coefficients(GFCC) is given out based on Gammatone Filter bank models.As the GFCC only reflects the static properties,the Gammatone Filter Shifted Delta Cepstral Coefficients(GFSDCC) is extracted based on Shifted Delta Cepstral.Then,the adaptive process for each GMM model with sufficient training data is transformed to the shift factor based on factor analysis.Furthermore,when the training data are insufficient,the coordinate of the shift factor is learned from the GMM mixtures of insensitive to the training data and then it is adapted to compensate other GMM mixtures.The experiment result shows that the recognition rate of the method proposed is 98.46%.The conclusion is that the performance of speaker recognition system is improved under several kinds of noise environments.

Research on Robust Speaker Identification Based on Adaptive Histogram Equalization

Diversified methods of decreasing the influence of noise have appeared to improve the performance of speaker recognition system in noise.In the paper,based on the statistical characteristics of speaker feature and the particularity of histogram equalization applying to speaker recognition,an adaptive histogram equalization method for speaker recognition is presented.The experiments showed improvements in performance with the proposed method in comparison with ordinary histogram equalization,and that the robustness of the system could be improved with the method under different noise environments.