Audio-Textual Emotion Recognition Based on Improved Neural Networks

Table 2

The comparison of different methods.

Method	WA	UA
CNN-LSTM + CNN (Gu et al.) [20]	60.40	59.88
RSM + SVM (Shah et al.) [31]	—	61.96
Our M_D_BL	65.59	67.76
Our M_CL_BL	66.26	67.25
Our M_CL_BLR	70.18	70.50
Our M_CBLA_BL	69.37	71.50
Our M_CBLA_BLR	70.40	71.25