《计算机应用研究》|Application Research of Computers

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

免费全文下载 (已被下载 次)  
获取PDF全文
作者 黄英来,李晓霜,赵鹏
机构 东北林业大学 信息与计算机工程学院,哈尔滨 150040
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2019)03-027-0776-05
DOI 10.19734/j.issn.1001-3695.2017.10.0990
摘要 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ReLU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。
关键词 卷积神经网络;网格搜索;语谱图;木材振动信号
基金项目 国家自然科学基金资助项目(31670717)
国家教育部新世纪优秀人才专项基金资助项目(NCET-12-0809)
中央高校基本科研业务费专项基金资助项目(2572018BH03)
本文URL http://www.arocmag.com/article/01-2019-03-027.html
英文标题 Research on wood quality for musical instrument recognition using convolutional neural network
作者英文名 Huang Yinglai, Li Xiaoshuang, Zhao Peng
机构英文名 CollegeofInformation&ComputerEngineering,NortheastForestryUniversity,Harbin150040,China
英文摘要 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 conve-nient method for the selection of the national musical instrument manufacturing field.
英文关键词 convolutional neural network; grid search; spectrogram; wood vibration signal
参考文献 查看稿件参考文献
  [1] 罗嘉琪, 唐衡. 中国民族乐器传承与发展[J] . 前沿, 2013, 30(7):166-167. (Luo Jiaqi, Tang Heng. The frontier and development of Chinese national musical instruments[J] . Forward Position, 2013, 30(7):166- 167. )
[2] 李源哲, 李先泽, 汪溪泉, 等. 几种木材声学性质的测定[J] . 林业科学, 1962, 7(1):59-66. (Li Yuanzhe, Li Xianze, Wang Xiquan, et al. Determination of acoustic properties of several kinds of wood[J] . Forestry Science, 1962, 7(1):59-66. )
[3] 沈隽, 刘一星, 刘振波, 等. 纤丝角对云杉属木材声振动特性的影响[J] . 东北林业大学学报, 2002, 30(5):50-52. (Shen Jun, Liu Yixing, Liu Zhenbo, et al. Effect of filaments angle on acoustic vibration characteristics of picea wood[J] . Journal of Northeast Forestry University, 2002, 30(5):50-52. )
[4] Ono T, Norimoto M. Study on Young’s modulus and internal friction of wood in relation to the evaluation of wood for musical instruments[J] . Limnology & Oceanography, 1983, 22(4):611-614.
[5] Tonosaki M, Okano T, Asano I. Vibrational properties of Sitka Spruce with longitudinal vibration and flexural vibration[J] . Mokuzai Gakkaishi, 1983, 29(9):547-552.
[6] Sobue N. Measurement of Young’s modulus by the transient longitudinal vibration of wooden beams using a fast Fourier transformation spectrum analyzer[J] . Journal of the Japan Wood Research Society, 1986, 32(9):744-747.
[7] Treu A, Hapla F. Study on the quality of spruce and fir resonance wood[J] . Allgemeine Forst-und Jagdzeitung, 2000, 171(12):215-222.
[8] LeCun Y, Bengio Y, Hinton G. Deep learning[J] . Nature, 2015, 521(7553):436-444.
[9] 周飞燕, 金林鹏, 董军. 卷积神经网络研究综述[J] . 计算机学报, 2017, 40(6):1229-1251. (Zhou Feiyan, Jin Linpeng, Dong Jun. Summary of convolution neural network research[J] . Acta Computer Science, 2017, 40(6):1229-1251. )
[10] Hinton G E, Salakhutdinov R R. Reducing the dimensionality of data with neural network[J] . Science, 2006, 313(5786):504-507. [11] Dennis J W. Sound event recognition in unstructured environments using spectrogram image processing[D] . Singapore:Nanyang Technological University, 2014.
[12] Dennis J, Tran H D. Enhanced local feature approach for overlapping sound event recognition[C] //Proc of Summit and Conference on Asia-Pacific Signal and Information Processing Association. Piscataway, NJ:IEEE Press, 2015:1-4.
[13] Zue V. Notes on speech spectrogram reading[M] . Cambridge, MA:MIT Press, 1991.
[14] Sainath T N, Mohamed A R, Kingsbury B, et al. Deep convolutional neural networks for LVCSR[C] //Proc of IEEE International Conference on Acoustics, Speech and Signal Processing. Piscataway, NJ:IEEE Press, 2013:8614-8618.
[15] Abdel-Hamid O, Mohamed A R, Jiang H, et al. Convolutional neural networks for speech recognition[J] . IEEE/ACM Trans on Audio Speech & Language Processing, 2014, 22(10):1533-1545.
[16] 邓柳, 汪子杰. 基于深度卷积神经网络的车型识别研究[J] . 计算机应用研究, 2016, 33(3):930-932. (Deng Liu, Wang Zijie. Research on vehicle type recognition based on deep convolution neural network[J] . Application Research of Computers, 2016, 33(3):930-932. )
[17] 李坚. 木材科学[M] . 3版. 北京:科学出版社, 2014. (Li Jian. Wood science[M] . 3rd ed. Beijing:Science Press, 2014. )
[18] 梁士利, 魏莹, 潘迪, 等. 基于语谱图行投影的特定人二字汉语词汇识别[J] . 吉林大学学报:工学版, 2017, 47(1):294-300. (Liang Shili, Wei Ying, Pan Di, et al. Two person word recognition based on line projection in Chinese[J] . Journal of Jilin University:Engineering Edition, 2017, 47(1):294-300. )
[19] 陶华伟, 査诚, 梁瑞宇, 等. 面向语音情感识别的语谱图特征提取算法[J] . 东南大学学报:自然科学版, 2015, 45(5):817-821. (Tao Huawei, Sui Cheng, Liang Ruiyu, et al. Spectrogram feature extraction algorithm for speech emotion recognition[J] . Journal of Southeast University:Natural Science, 2015, 45(5):817-821. )
[20] LeCun Y. LeNet-5, convolutional neural networks[EB/OL] . (2015). http://yann. lecun. com/exdb/lenet.
[21] Srivastava N, Hinton G E, Krizhevsky A, et al. Dropout:a simple way to prevent neural networks from overfitting[J] . Journal of Machine Learning Research, 2014, 15(1):1929-1958.
[22] Hinton G E, Srivastava N, Krizhevsky A, et al. Improving neural networks by preventing co-adaptation of feature detectors[J] . Computer Science, 2012, 3(4):212-223.
[23] https://keras. io/[EB/OL] .
[24] 李瀚, 杨晓峰, 邓红霞, 等. 基于网格搜索算法的PCNN模型参数自适应[J] . 计算机工程与设计, 2017, 38(1):192-197. (Li Han, Yang Xiaofeng, Deng Hongxia, et al. Parameter adaptation of PCNN model based on grid search algorithm[J] . Computer Engineering and Design, 2017, 38(1):192-197. )
收稿日期 2017/10/31
修回日期 2017/12/15
页码 776-780
中图分类号 TP391
文献标志码 A