基于 matlab 的语音信号处理
摘 要:对录制的语音信号进行采样,分析其时域波形和频谱图。给定数字滤波器的性能指
标,采用窗函数法和双线性变换法设计数字滤波器,并对语音信号进行滤波,得到滤波前后的
信号幅频响应。通过对比高通、低通两种滤波处理结果,简单而有效地论证了两种数字滤波
器在语音信号处理上的异同。并进一步求出特征频段语音信号随时间变化的曲线,分析自
身语音信号的特点。
关键词: MATLAB 数字滤波器 语音信号
Speech Signal Processing by Digital Filter Based on MATLAB
Abstract :Time-domain waveform and frequency spectrum of the recorded speech signals are
analyzed by sampling. The performance indexes of digital filters are given. Two methods of
window function and bilinear transformation are used to design the digital filters. The speech
signal is filtered by the filters, and then magnitude-frequency responses of the signal before and
after filtering are received. The advantages of two digital filters (filter low pass and filter high
pass)in speech signal processing are demonstrated by comparing different methods for filtering
simply and effectively. For more, we are able to figure out the time curves of characteristic bands
of speech signal and then analyses the character of our own speech signals.
Key words: MATLAB, digital filter, speech signal
为了进一步观察和确定语音信号的频谱特征,首先分别画出每秒的频带特征。具体代码
如下所示:
clear
[yfull,fs,nbits] = wavread('D:\meng');
y = yfull(:,1);
win = blackman(fs);
yfft = zeros(15,fs);
ypower = zeros(15,fs);
for ii = 0:14
yfft(ii+1,1:fs) = abs(fft(y(1+ii*fs:(ii+1)*fs).*win,fs));
ypower(ii+1,1:fs) = yfft(ii+1,1:fs).^2;
figure(ii+1);
f = 1:3000;5
plot(f,ypower(ii+1,1:3000));
end
由此得到十五秒中每一秒的强度分布(注:横坐标为频率,纵坐标为强度)
评论0
最新资源