Speaker age estimation using Hidden Markov Model weight supervectors

This paper proposes a new approach for speaker age estimation. In this method, speakers are modeled by their corresponding Hidden Markov Model (HMM) weight supervectors. Then, Weighted Supervised Non-Negative Matrix Factorization (WSNMF) is applied to reduce the dimension of the input space. Finally, a Least Squares Support Vector Regressor (LS-SVR) is employed to estimate the age of speakers using the obtained low-dimensional vectors. Evaluation results on a corpus of read and spontaneous speech in Dutch confirms the effectiveness of the proposed scheme.