卷积神经网络在乐器板材优劣识别中的应用研究

Research on wood quality for musical instrument recognition using convolutional neural network

目前民族乐器板材振动信号识别算法存在特征提取复杂且耗时长等缺点,针对此问题,提出了一种基于卷积神经网络的木材振动信号分类识别算法,实现了乐器板材优劣的判别。卷积神经网络将特征提取和分类过程结合来进行神经网络的训练,具有识别度高、鲁棒性好等优点。首先重点分析和讨论了提取木材振动信号的语谱图特征,然后应用卷积神经网络结合网格搜索的方法进行参数调优。为了防止过拟合,还应用了Re LU和dropout等新技术,得到最终分类结果。实验证明,测试样本准确率达到96%,明显优于传统方法。该方法可减小人工测量的误差,加快板材的选取时间,为民族乐器制造领域的选材提供了一种更加实用的方法。

At present,the vibration signal recognition algorithm for national musical instrument plate has the shortcomings of complex feature extraction and time-consuming.To solve this problem,this paper proposed a classification algorithm of wood vibration signal based on convolution neural network,to identify the quality of the musical instrument.Convolution neural network combined feature extraction and classification process to train the neural network,which owned the advantages of high recognition rate and good robustness.Firstly,this paper mainly analyzed and discussed spectrogram characteristics of the extraction of wood vibration signals.Then combining convolution neural network and grid search method,it could adjust the parameters.In order to avoid over-fitting,it obtained the final classification results by using new technologies such as ReLU and dropout.The experiments show that the accuracy of the test sample reaches 96%,which is obviously better than the traditional method.This method can reduce the error of manual measurement and speed up the selection time of the plate,and provide a more convenient method for the selection of the national musical instrument manufacturing field.