Computer Engineering and Applications ›› 2008, Vol. 44 ›› Issue (18): 246-248.

• 工程与应用 • Previous Articles    

Consecutive humming feather extraction based on two-layer neural network

ZHENG Gui-bin1,LIU Yan2,LIU Sheng1,HAN Ji-qing3   

  1. 1.College of Automation,Harbin Engineering University,Harbin 150001,China
    2.College of Computer Science and Technology,Harbin University of Science and Technology,Harbin 150080,China
    3.College of Computer Science and Technology,Harbin Institute of Technology,Harbin 150001,China
  • Received:2007-09-24 Revised:2007-11-30 Online:2008-06-21 Published:2008-06-21
  • Contact: ZHENG Gui-bin

基于两级神经网络的连续哼唱特征提取

郑贵滨1,刘 艳2,刘 胜1,韩纪庆3   

  1. 1.哈尔滨工程大学 自动化学院,哈尔滨 150001
    2.哈尔滨理工大学 计算机科学与技术学院,哈尔滨 150080
    3.哈尔滨工业大学 计算机科学与技术学院,哈尔滨 150001
  • 通讯作者: 郑贵滨

Abstract: A two-layer neural network for humming feature extraction is proposed for note segmentation and note recognition,when users sing consecutively with arbitrary words.The first-layer BP network is used to divide the humming data into the independent notes.The second-layer RBF network is used to receives the MIDI pitch of the independent note.The experimental results show that this method can extract the humming features with high accuracy,and it’s suitable to the QBH(Query By Humming) system.

摘要: 针对用户以任意字词连续哼唱的情况下,哼唱特征提取中音符分割、音符识别难度大的问题,提出了一种基于两级神经网络的哼唱特征提取方法。第一级采用BP神经网络实现哼唱音符分割,得到独立音符;第二级采用RBF神经网络识别分割出的各个音符,获得音符的MIDI音高值。实验结果表明,该方法能较好地完成哼唱特征的提取,适合于实际哼唱检索系统对连续哼唱的要求。