基于反向传播算法与隐藏层的语音识别系统研究

Research on Speech Recognition System Based on Backpropagation Algorithm and Hidden Layer

为设计出具有优秀分类性能的英语语音情感识别模型,研究中调整了反向传播算法的隐层结构,并与海鸥优化算法结合组成英语语音情感识别模型。测试结果如下:设计的模型在训练完成后损失函数为1.33,低于对比模型,它的识别准确率与精确率分别为96.1%与96.5%,也高于全部对比模型。测试结果证明,此次研究设计的英语语音情感识别模型具有优良的英语语音识别能力,在提升英语机器翻译系统翻译能力方面,具有一定应用潜力。

In order to design an English speech emotion recognition model with excellent classification performance,the hidden layer structure of the back-propagation algorithm is adjusted in the research,and the English speech emotion recognition model is composed of the combination of the back-propagation algorithm and the seagull optimization algorithm.The test results are as follows:The designed model has a loss function of 1.33 after training,which is lower than the comparison model.Its recognition accuracy and accuracy are 96.1%and 96.5%,respectively,which are also higher than all the comparison models.The test results prove that the English speech emotion recognition model designed in this study has excellent English speech recognition ability,and has certain application potential in improving the translation ability of English machine translation systems.