国际中文课堂中的教师语音情感识别研究

Research on Teacher Speech Emotion Recognition in International Chinese Language Classrooms

当前,国际中文教育正处于转型发展的关键期,亟待引入人工智能、大数据等新一代信息技术以提高教育教学质量,推动中文教育数字化、智慧化。为此,文章首先以通道注意力机制、时间注意力机制以及空间注意力机制为基础,融合2D、3D卷积神经网络,构建多维注意力混合卷积神经网络模型用于语音情感识别。然后,文章将该模型应用于国际中文课堂教师语音情感识别实践,发现该模型通过深度学习能够更好地获取语音情感特征表示。此外,文章还发现模型表现出较好的语音特征提取和分类能力,能够对国际中文课堂中教师普遍表现的高兴、关怀、满意、平静等情感进行识别。文章通过研究,旨在助益中文教师自主提升教学实践能力、改善课堂教学效果。

At present,international Chinese language education is in a critical period of transformation and development,and it is urgent to introduce the new generation of information technology of artificial intelligence and big data to improve the quality of education and teaching,and futher promote the digitalization and intelligence of Chinese language education.In this paper,based on the channel attention mechanism,temporal attention mechanism and spatial attention mechanism and fused 2D and 3D convolutional neural networks,a multidimensional attention hybrid convolutional neural network model was built for speech emotion recognition.Then the model was applied to the teacher speech emotion recognition practice in international Chinese language classrooms.And it was found that the model can better obtain speech emotion feature representation through deep learning.In addition,the model exhibited good speech feature extraction and classification capabilities,and could recognize the emotions of joy,care,satisfaction,and calmness that were commonly expressed by teachers in international Chinese language classrooms.Through the study,this paper was aimed to help Chinese teachers improve their teaching practices and improve the teaching effectiveness in classrooms.