声音的数学表示和变换.pdf

声音的数学表示和变换 清华大学附属中学 陈嘉雯 指导教师：李劲松 摘要：人耳可以听到各种各样的声音。为什么不同乐器演奏同一音符，声音却很 不相同？为什么不同人用同样的速度和音调唱同一首歌或朗读同一段课文，声音 也如此不同？本文首先介绍了声音在时域和频域的数学表示，并借助MATLAB ， 利用离散Fourier 变换和反变换，在两种表示形式之间进行转换。接着尝试分别 从时域和频域，对声音进行变换。在时域，通过对声音信号进行时间轴压缩或拉 伸，变换后的声音，音调和音色将显著改变。在频域，通过对频谱进行滤波处理， 对某些频率点/频率段的频谱强度进行调整、或者将频谱整体平移，可以变换得 到与原有音色不同的声音。最后，在MATLAB 环境中，实现了一个简易的声音 变换应用。通过图形交互界面，可以从时域和频域对钢琴音符、唱歌片段和朗读 片段进行声音变换。 关键词：声音，离散Fourier 变换/反变换，MATLAB ，频谱强度，滤波 Mathematical Representation and Transformation of Sound Abstract: Human beings can hear a variety of sounds. Why does it sound so different from each other when a same musical note is played by different instruments and a same song or paragraph is sung or read by different people even at the same speed and in the similar tune? Firstly mathematical representations of sound, in both time and frequency domain, are introduced. By use of DFT (Discrete Fourier Transform) and IDFT (Inverse Discrete Fourier Transform), transformation between these two representations is demonstrated in MATLAB. After that, ways of sound transformation are explored from both time and frequency domain. In time domain, when sound signal is compressed or stretched in time axis, both tune and tone are significantly changed. While in frequency domain, sound in different tones can be acquired through spectrum filtering or intensity altering on some certain frequency points or segments. Finally, a preliminary sound transformation application is built via MATLAB. Through a graphical user interface, time-domain and frequency-domain sound transformation result could be carried out for a piece of piano note, song segment and reading paragraph. Keywords ：Sound, Discrete Fourier Transform, Inverse Discrete Fourier Transform, MATLAB, Spectrum intensity, Filtering 1 1. 概述 1.1 声音 声音是物体振动产生的声波。人说话或唱歌时，声带振动发出声音。钢琴的 琴键被