基于TMS320VC5509-DSP的語音信號的FIR濾波器

1、2009級學生課程設計 DSP原理與應用 課程設計報告書課題名稱基于TMS320VC5509 DSP的語音信號的FIR濾波器姓 名學 號 院、系、部專 業(yè)指導教師 20012年 6月 9日 基于TMS320VC5509 DSP的語音信號的FIR濾波器設計一、實踐的目的和要求 1.熟悉ICETEK-VC5509-A板上語音codec芯片TLV320AIC23的設計和程序控制原理。2.熟悉FIR濾波器工作原理及編程。3.掌握使用TI的算法庫dsplib提高程序運行效率的方法。4.學習使用CCS圖形觀察窗口觀察和分析語音波形及其頻譜。二、實踐原理： 在數(shù)字信號處理中，濾波占有極其重要的地位。數(shù)字濾波

2、是語音處理、圖像處理、模式識別、頻譜分析等應用中的基本處理算法。用DSP芯片實現(xiàn)數(shù)字濾波除了具有穩(wěn)定性好、精確度高、不受環(huán)境影響等優(yōu)點外，還具有靈活性好等特點。 VC5509引腳圖過渡帶寬度=阻帶邊緣頻寬-通帶邊緣頻率采樣頻率：f1=通帶邊緣頻率+（過渡帶寬度）/2理想低通濾波器脈沖響應：h1n=sin(n1)/n/根據(jù)要求，選擇布萊克曼窗，窗函數(shù)長度為：N=5.98fs/過渡帶寬度選擇N=30wn=0.42+0.5cos(2n/30)+0.8cos(4n/24)濾波器脈沖響應為：hn=h1nwn|n|12根據(jù)上面計算，各式算出hn,然后將脈沖響應值移位為因果序列。完成的濾波器的差分方程為：y

3、n=-0.001xn-2-0.002xn-3-0.002xn-4+0.01xn-5-0.009xn-6-0.018xn-7-0.049xn-8-0.02xn-9+0.11xn-10+0.28xn-11+0.64xn-12+0.28xn-13-0.11xn-14-0.02xn-15+0.049xn-16-0.018xn-17-0.009xn-18+0.01n-19-0.002xn-20-0.002xn-21+0.001xn-22開始初始化：EMIF、CPU頻率、AIC23調用AIC23_Mixer子程序處理音頻數(shù)據(jù)輸入/輸出開始初始化語音緩沖區(qū)和工作變量等待McBSP通道0傳送結束讀取語音數(shù)據(jù)保

4、存于左聲道緩沖區(qū)調用fir2計算FIR濾波輸出濾波結果保存于右聲道緩沖區(qū)原聲音送左聲道，濾波結果送右聲道，輸出到McBSP0三、實踐步驟1.首先自己參照指導書完語音信號的FIR濾波實驗，認真閱讀實驗中的源程序，深刻理解語音信號FIR濾波的原理及具體實現(xiàn)方法，包括含噪語音信號的讀取，濾波后信號的輸出，語音編解碼器的設置（AIC23），重點理解FIR濾波器的實現(xiàn)（循環(huán)尋找的實現(xiàn)）2.語音采集的采集與分析 （1）錄制語音，并命名為“e”，保存在相應的位置（桌面）。（2）用MATLAB對原始語音信號進行分析，畫出它的時域波形和頻譜。程序如下：fs=22050;x,fs,Nbits=wavread(C:

5、Documents and SettingsOwner桌面e);t=0:1/22050:(length(x)-1)/22050;sound(x,fs) ; %對加載的語音信號進行回放figure(1)plot(x) %做原始語音信號的時域圖形title(原始語音信號);xlabel(time n);ylabel(fuzhi n);figure(2)freqz(x) %繪制原始語音信號的頻率響應圖title(頻率響應圖)n=length(x); %求出語音信號的長度y1=fft(x,n) ; %傅里葉變換y2=fftshift(y1); %對頻譜圖進行平移f=0:fs/n:fs*(n-1)/n;

6、 %得出頻點figure(3) subplot(2,1,1);plot(abs(y2) %做原始語音信號的FFT頻譜圖title(原始語音信號FFT頻譜)subplot(2,1,2);plot(f,abs(y2); %繪制原始語音信號的頻譜圖title(原始語音信號頻譜)%加噪聲k=1:1:n; %定義k 值,噪聲與原始語音長度一致c1=100000*sin(2*pi*5000*k); %噪聲信號的函數(shù)z=x+c1; %噪音與原始信號疊加wavwrite(z,8000,16,C:Documents and SettingsOwner桌面e2); %得到含噪語音s=length(c1);y3=f

7、ft(z,s); %對疊加信號進行頻譜變換y4=fftshift(y3); %對頻譜圖平移f=0:fs/s:fs*(s-1)/s; %得出頻點figure(4)plot(z) %做原始語音信號的時域圖形title(加噪語音信號);xlabel(time n);ylabel(fuzhi n);figure(5)plot(f,abs(y4); %繪出加噪語音頻譜圖axis(0 8000 0 100);title(含噪語音信號的頻譜);figure(6)subplot(2,1,1);plot(f,abs(y2) %做原始語音信號的FFT頻譜圖title(原始語音信號FFT頻譜)subplot(2,1

8、,2);plot(f,abs(y4) %做加噪語音信號的FFT頻譜圖title(加噪語音信號頻譜) 原始語音信號 加噪語音信號 原始語音信號FFT頻譜 含噪語音信號頻譜 原始語音信號FFT頻譜 頻率響應圖3. 設計一定參數(shù)的濾波器使用fir2函數(shù)設計一個31階低通濾波器，信號的采樣頻率為8000Hz，濾波器的通帶截止頻率為2900Hz，阻帶截止頻率為3300Hz。在函數(shù)中，其截止頻率均用歸一化頻率表示。歸一化頻率的計算方法：f=實際頻率/(采樣頻率/2)，則將截止頻率歸一化后的截止頻率值為0.725、0.825用fir2函數(shù)實現(xiàn)的格式：f = 0 0.725 0.825 1 ； m = 1 1

9、 0 0 ； b=fir2（30，f，m）；b= -0.0005,0.0004,0.0003,-0.0018,0.0039,-0.0045,0.0009,0.0086,-0.0210,0.0279,-0.0173,-0.0191,0.0793,-0.1486,0.2043,0.7744,0.2043,-0.1486,0.0793,-0.0191,-0.0173,0.0279,-0.0210,0.0086,0.0009,-0.0045,0.0039,-0.0018,0.0003,0.0004,-0.0005Audio濾波程序如下：#include myapp.h#include ICETEK-V

10、C5509-EDU.h#include scancode.h#include #define FIRNUMBER 31#define SIGNAL1F 2900#define SIGNAL2F 3300#define SAMPLEF 8000#define PI 3.1415926float InputWave();float FIR();float fHnFIRNUMBER=-0.0005,0.0004,0.0003,-0.0018,0.0039,-0.0045,0.0009,0.0086,-0.0210,0.0279,-0.0173,-0.0191,0.0793,-0.1486,0.204

11、3,0.7744,0.2043,-0.1486,0.0793,-0.0191,-0.0173,0.0279,-0.0210,0.0086,0.0009,-0.0045,0.0039,-0.0018,0.0003,0.0004,-0.0005;float fXnFIRNUMBER= 0.0 ;float fInput,fOutput;float fSignal1,fSignal2;float fStepSignal1,fStepSignal2;float f2PI;int i;float fIn256,fOut256;int nIn,nOut;main()nIn=0; nOut=0;f2PI=2

12、*PI;fSignal1=0.0;fSignal2=PI*0.1;fStepSignal1=2*PI/30;fStepSignal2=2*PI*1.4;while ( 1 )fInput=InputWave();fInnIn=fInput;nIn+; nIn%=256;fOutput=FIR();fOutnOut=fOutput;nOut+;/* break point */if ( nOut=256 )nOut=0;float InputWave()for ( i=FIRNUMBER-1;i0;i- )fXni=fXni-1;fXn0=sin(double)fSignal1)+cos(dou

13、ble)fSignal2)/6.0;fSignal1+=fStepSignal1; if ( fSignal1=f2PI )fSignal1-=f2PI;fSignal2+=fStepSignal2;if ( fSignal2=f2PI )fSignal2-=f2PI;return(fXn0);float FIR()float fSum;fSum=0;for ( i=0;i spcr1, 0); Write(pMCBSP0 - spcr2, 0); / Config frame parameters (32 bit, single phase, no delay) Write(pMCBSP0

14、- xcr1, XWDLEN1_32); Write(pMCBSP0 - xcr2, XPHASE_SINGLE | XDATDLY_0); Write(pMCBSP0 - rcr1, RWDLEN1_32); Write(pMCBSP0 - rcr2, RPHASE_SINGLE | RDATDLY_0); / Disable int frame generation and enable slave w/ext frame signals on FSX / Frame sync is active high, data clocked on rising edge of clkx Writ

15、e(pMCBSP0 - pcr, PCR_CLKXP); / Bring transmitter and receiver out of reset SetMask(pMCBSP0 - spcr2, SPCR2_XRST); SetMask(pMCBSP0 - spcr1, SPCR1_RRST); void AIC23_Init() I2C_Init(); / Reset the AIC23 and turn on all power AIC23_Write(AIC23_RESET_REG, 0); AIC23_Write(AIC23_POWER_DOWN_CTL, 0); AIC23_Wr

16、ite(AIC23_ANALOG_AUDIO_CTL, ANAPCTL_DAC | ANAPCTL_INSEL);/ 使用麥克風音源 AIC23_Write(AIC23_DIGITAL_AUDIO_CTL, 0); / Turn on volume for line inputs AIC23_Write(AIC23_LT_LINE_CTL,0x000); AIC23_Write(AIC23_RT_LINE_CTL,0x000); / Configure the AIC23 for master mode, 44.1KHz stereo, 16 bit samples / Use 12MHz U

17、SB clock AIC23_Write(AIC23_DIGITAL_IF_FORMAT, DIGIF_FMT_MS | DIGIF_FMT_IWL_16 | DIGIF_FMT_FOR_DSP); AIC23_Write(AIC23_SAMPLE_RATE_CTL, SRC_SR_44 | SRC_BOSR | SRC_MO); / Turn on headphone volume and digital interface AIC23_Write(AIC23_LT_HP_CTL, 0x07f); / 0x79 for speakers AIC23_Write(AIC23_RT_HP_CTL

18、, 0x07f); AIC23_Write(AIC23_DIG_IF_ACTIVATE, DIGIFACT_ACT); / Set McBSP0 to be transmit slave McBSP0_InitSlave();void AIC23_Disable() PC55XX_MCSP pMCBSP0 = (PC55XX_MCSP)C55XX_MSP0_ADDR; I2C_Disable(); / Put the MCBSP in reset Write(pMCBSP0 - spcr1, 0); Write(pMCBSP0 - spcr2, 0);#define AUTIODATALEFT

19、 0x0d000#define AUTIODATARIGHT 0x17000int *pAudioLeft,*pAudioRight;int www=0; void AIC23_Mixer() PC55XX_MCSP pMCBSP0 = (PC55XX_MCSP)C55XX_MSP0_ADDR; int left, right; int *pl,*pr,nAudioCount; int i;pAudioLeft=pl=(int *)AUTIODATALEFT;pAudioRight=pr=(int *)AUTIODATARIGHT;nAudioCount=0; for ( i=0;iNX;i+

20、) xi=0; for ( i=0;i spcr2, SPCR2_XRDY);/ 等待數(shù)據(jù)傳輸完成 (*pl)=left = Read(pMCBSP0 - ddr1);/ 讀入左聲道數(shù)據(jù) right = Read(pMCBSP0 - ddr2);/ 讀入右聲道數(shù)據(jù) xNX-1=left/16;/ 防止濾波時數(shù)據(jù)溢出 fir2(x, h, r, db, NX, NH);/ 調用濾波程序計算當前輸出(*pr)=rNX-1;/ 數(shù)組r的最后一個單元為當前輸出 Write(pMCBSP0 - dxr1, left);/ 將原始數(shù)據(jù)送左聲道輸出 /Write(pMCBSP0 - dxr1, rNX-1);/ 將