Title :
Group sparse features for speech emotion perception in tensor space
Author :
Qiang Wu ; Ju Liu ; Jiande Sun ; Jie Li ; Liqing Zhang
Author_Institution :
Sch. of Inf. Sci. & Eng., Shandong Univ., Jinan, China
Abstract :
With increasing demands for a natural interaction between human and machine, emotion perception from speech signals is becoming an important interaction interface. In this paper, we give a feature extraction framework for speech emotion recognition and present a novel method to extract emotion information based on group sparsity in tensor space. The speech signal is encoded as cortical representation in auditory system. We propose the group lasso nonnegative tensor factorization model to learn the multilinear factor matrices from tensor feature subspaces. l1/l2 constraint on multiple subspaces is imposed to recover the different groups of covariance for each factor (frequency, time, etc). The experimental results show that the proposed method can improve the multi-classes emotion recognition performance than state of the art baseline systems.
Keywords :
covariance matrices; emotion recognition; encoding; feature extraction; matrix decomposition; optimisation; sparse matrices; speech recognition; tensors; auditory system; cortical representation; covariance groups; emotion information extraction; feature extraction framework; group lasso nonnegative tensor factorization model; group sparse features; human-machine interaction; interaction interface; l1/l2 constraint; multiclass emotion recognition performance improvement; multilinear factor matrix learning; speech emotion perception; speech emotion recognition; speech signal encoding; speech signals; tensor feature subspaces; tensor space; Accuracy; Auditory system; Emotion recognition; Feature extraction; Speech; Speech recognition; Tensile stress;
Conference_Titel :
Mechatronics and Control (ICMC), 2014 International Conference on
Print_ISBN :
978-1-4799-2537-7
DOI :
10.1109/ICMC.2014.7231570