Multitask speaker profiling for estimating age, height, weight and smoking habits from spontaneous telephone speech signals

Abstract

This paper proposes a novel approach for automatic estimation of four important traits of speakers, namely age, height, weight and smoking habit, from speech signals. In this method, each utterance is modeled using the i-vector framework which is based on the factor analysis on Gaussian Mixture Model (GMM) mean supervectors, and the Non-negative Factor Analysis (NFA) framework which is based on a constrained factor analysis on GMM weights. Then, Artificial Neural Networks (ANNs) and Least Squares Support Vector Regression (LSSVR) are employed to estimate age, height and weight of speakers from given utterances, and ANNs and logistic regression (LR) are utilized to perform smoking habit detection. Since GMM weights provide complementary information to GMM means, a score-level fusion of the i-vector-based and the NFA-based recognizers is considered for age and smoking habit estimation tasks to improve the performance. In addition, a multitask speaker profiling approach is proposed to evaluate the correlated tasks simultaneously and in interaction with each other, and consequently, to boost the accuracy in speaker age, height, weight and smoking habit estimations. To this end, a hybrid architecture involving the score-level fusion of the i-vector-based and the NFA-based recognizers is proposed to exploit the available information in both Gaussian means and Gaussian weights. ANNs are then employed to share the learned information with all tasks while they are learned in parallel. The proposed method is evaluated on telephone speech signals of National Institute for Standards and Technology (NIST) 2008 and 2010 Speaker Recognition Evaluation (SRE) corpora. Experimental results over 1194 utterances show the effectiveness of the proposed method in automatic speaker profiling.