Final Project FAN WEN Introduction Goal To analyze a voice that is before adding noise, after adding noise and after filtered. Filter FIR filter- Finite impulse response filter. FIR filter having the following characteristics: (1). Phase can be made strictly linear, and they can have an arbitrary amplitude characteristics. (2). Unit impulse response is finite length, so it must be stable. FIR filter unit impulse response is h(n), 0≤n≤N-1. System functions: The form of differential equations: IIR filter-Infinite impulse response filter. IIR digital filter is an infinite impulse response sequence of units. Amplitude-frequency characteristics of IIR digital filter has high precision, it is not linear phase. Application Step1 Load a voice in the MATLAB. Making WAV files transform to variable. Calculate the array length. (I.e. it is the greater one between number of rows and columns) Find the spectrum of original signal .(using FFT) Draw the original signal waveform and spectrum with MATLAB. Step2 Define a random noise.(using function randn(n,1)) Adding the noise into original voice. Find the spectrum of the signal which is adding the noise.(using FFT) Draw the signal that is already added the noise waveform and spectrum with MATLAB. Step3 Design a low pass filter. Parameter of filter: Window: hamming. Order: 16 Fs: 8000 Fc: 1200 Let the mixed signal through the filter. Find the spectrum of the signal which is already filtered. Draw the signal waveform and spectrum after filtered with MATLAB. Step4 Replay and listen the voices, to see what differences with those voices. To set out those new generation of audio files. Result The filter can filter high frequency noise, but cannot clean all noise that added into the signal. Because it is a low pass filter, we find the voice getting dreary and low. Signal waveforms and spectrums: Conclusion In this project, I select a speech signal as the analysis object, and carries on the spectrum analysis. Using random function of MATLAB to create noise and add the noise to the voice signal, to imitate the noise speech signal, and carries on the spectrum analysis. Design FIR filter with MATLBA, and carries on speech signal contaminated by noise filtering, analysis the signal features of time domain and frequency domain which is after filtering. Reply the speech signal. In summary, the project is successful. I have learned that theory to practical combination is very important. I know how to use MATLAB and many knowledge of filter and signal by this project. Bibliography & References [1].CHI-TSONG C. Digital Signal Processing-Spectral Computation and Filter Design [M].London Oxford University Press,2002:198-205. [2]. John G. Proakis, Dimitris G Manolakis. Digital Signal Processing 4th edition Upper Saddle River, N.J.: Pearson Prentice Hall, 2007. MATLAB CODE [y,fs,bits]=wavread('test'); n=length(y); Y=fft(y); figure subplot(2,1,1);plot(y); title('The original signal waveform');grid; subplot(2,1,2);plot(abs(Y)); title('Spectrum of the original signal');grid; sound(y); Noise=0.05*randn(n,1); s=y+Noise; S=fft(s); figure subplot(2,1,1);plot(s); title('signal waveform(After adding noise)');grid; subplot(2,1,2);plot(abs(S)); title('signal spectrum(After adding noise)');grid; sound(s); a=GS;b=1;c=filter(a,b,s); C=fft(c); figure subplot(2,1,1);plot(c); title('signal waveform(After filter noise)');grid; subplot(2,1,2);plot(abs(C)); title('signal spectrum(After filter noise)');grid; sound(c); wavwrite(s,fs,bits,'test+Noise.wav'); wavwrite(c,fs,bits,'test+Noise+filter.wav');