Analyzing the Relationship between Students' Emotional Changes and Ideological and Political Quality Enhancement in the Process of Civic Education Using LSTM Network Modeling

Abstract

In this paper, we first preprocess the speech signal by methods such as pre-emphasis and windowing, and then extract the speech features by using the Mel Frequency Cepstrum Coefficients (MFCC) in the spectral features, and construct a speech emotion recognition model based on the bidirectional LSTM to capture the semantic correlation between long sequences. In order to synthesize the complementary information in the dataset and reduce feature redundancy, this paper constructs a speech emotion recognition model incorporating two-way CNN-LSTM and attention mechanism to analyze the emotional changes of students in the process of Civic Education. The results show that the method in this paper achieves the highest performance on the CASIA emotion corpus. The classroom as a whole presents a positive and active state (most of the emotion value P is distributed between [0,2]), the classroom atmosphere is good, and the teacher-student interaction has a positive effect on the improvement of ideological and political quality; in addition, the classroom emotion can also affect the classroom effect to a certain extent, which influences the effect of the students' quality improvement in the process of Civic and political education. Reasonably guiding the students' emotions such as “conscientiousness and empathy sublimation” in the classroom of Civic and Political Education can effectively improve the students' ideological and political quality.