基于时频分析的乐音识别算法研究-模式识别与智能系统专业论文.docx

中国科学技术大学硕：七学位论文 中国科学技术大学硕：七学位论文 英文摘要 Abstract Music signal processing is an important part of signal processing．With the popularization of computer,the digital multimedia entertainment and intelligent human-machine interface require that computers have friendly appearances in visual and audio．In order to communicate with computer freely’the research on computer music signals processing has been taking a more and more important role． First of all，we describe briefly about the development of Computer Music， including MIDI(Musical Instrument Digital Interface)，Computer Aided Composition etc．Then，we introduce the content of general music recognition such as rhythm， melody,structure genre recognition．This is decided by the complexity of music works： Music recognition includes lots of fields of modem signal processing and pattern recognition．We describe the music physics，zero-pole point model，hearing model and we emphasize the time—frequency analysis theory，short time Fourier Transform and Wavelet Transform．Their conception，proprieties are described in detail．Meanwhile we introduce some recognition strategies such HMM and ANN，we present some comparison and comments． We discuss the implementation of algorithm，including general programming，data processing．We describe three algorithms such as note recognition based on multiple flame statistics，music segmentation based on dynamic envelope detection and chords recognition based on template mating and decision trees with max likelihood coefficients，Compared the methods mentioned before，these algorithms are more effective and efficient．At last，we make a conclusion and discuss the future WOrk． Keywords：Music Recognition，Time—frequency Analysis，Wavelet Transform 中国科学技术大学硕士学位论文 中国科学技术大学硕士学位论文 第一章前言 第一章前言 1．1研究背景 1．1．1计算机音乐发