摘要
本文设计了一种基于CNN和LSTM语音情感识别神经网络结构,其在识别多分类情感时表现良好,与使用单卷积核、双卷积核网络结构相比,本文结构表现更加优秀。
In this paper,a neural network structure based on CNN and LSTM for speech emotion recognition is designed.It performs well in the recognition of multi-category emotions.Compared with the network structure using single convolution kernel and double convolution kernel,the structure of this paper performs better.
作者
梁晓鸽
赵风海
LIANG Xiaoge;ZHAO Fenghai(School of Electronic Information and Optical Engineering,Nankai University,Tianjin 300350,China)
出处
《电声技术》
2020年第3期27-28,36,共3页
Audio Engineering
