DSP实验答案Solutionlab5精.docx

DSP实验答案Solutionlab5精.docx

《DSP实验答案Solutionlab5精.docx》由会员分享，可在线阅读，更多相关《DSP实验答案Solutionlab5精.docx（12页珍藏版）》请在冰豆网上搜索。

DSP实验答案Solutionlab5精

Name:

Section:

LaboratoryExercise5

LINEAR,TIME-INVARIANTDISCRETE-TIMESYSTEMS:

FREQUENCY-DOMAINREPRESENTATIONS

5.1TRANSFERFUNCTIONANDFREQUENCYRESPONSE

Project5.1TransferFunctionAnalysis

Answers:

Q5.2TheplotofthefrequencyresponseofthecausalLTIdiscrete-timesystemofQuestionQ5.2obtainedusingthemodifiedprogramisgivenbelow:

Thetypeoffilterrepresentedbythistransferfunctionis-BF

5.2TYPESOFTRANSFERFUNCTIONS

Project5.2Filters

AccordingtoProgrammeP4_1.

Answers:

Q5.8TherequiredmodificationstoProgramP4_1tocomputeandplottheimpulseresponseoftheFIRlowpassfilterofProject4.2withalengthof20andacutofffrequencyofωc=0.45

areasindicatedbelow:

clf;

fc=0.45;

n=[-10:

1:

10];

y=2*fc*sinc（2*fc*n;k=n+10;

stem（k,y;title（'N=20';axis（[020-0.20.6];

xlabel（'Timeindexn';ylabel（'Amplitude';grid;

Theplotgeneratedbyrunningthemodifiedprogramisgivenbelow:

Q5.20TheplotsoftheimpulseresponsesofthefourFIRfiltersgeneratedbyrunningProgramP4_3aregivenbelow

:

Fromtheplotswemakethefollowingobservations:

Filter#1isoflength____9___witha__symmetrical____impulseresponseandisthereforeaType_1_linear-phaseFIRfilter.

Filter#2isoflength____10__witha__antisymmetric____impulseresponseandisthereforeaType_2_linear-phaseFIRfilter.

Filter#3isoflength___9_____witha___symmetrical____impulseresponseandisthereforeaType_3_linear-phaseFIRfilter.

Filter#4isoflength__10_____witha__antisymmetric___impulseresponseandisthereforeaType_4_linear-phaseFIRfilter.

FromthezerosofthesefiltersgeneratedbyProgramP4_3weobservethat:

Filter#1haszerosatz=2.3273+2.0140i2.3273-2.0140i

-1.2659+2.0135i-1.2659-2.0135i

-0.2238+0.3559i-0.2238-0.3559i

0.2457+0.2126i0.2457-0.2126i

Filter#2haszerosatz=2.5270+2.0392i2.5270-2.0392i

-1.0101+2.1930i-1.0101-2.1930i

-1.0000

-0.1733+0.3762i-0.1733-0.3762i

0.2397+0.1934i0.2397-0.1934i

Filter#3haszerosatz=-1.00000.2602+1.2263i

0.2602-1.2263i1.0000

0.6576+0.7534i0.6576-0.7534i

0.1655+0.7803i0.1655-0.7803i

Filter#4haszerosatz=2.0841+2.0565i2.0841-2.0565i

-1.5032+1.9960i-1.5032-1.9960i

1.0000

-0.2408+0.3197i-0.2408-0.3197i

0.2431+0.2399i0.2431-0.2399i

PlotsofthephaseresponseofeachofthesefiltersobtainedusingMATLABareshownbelow

:

Fromtheseplotsweconcludethateachofthesefiltershave____linear___phase.

ThegroupdelayofFilter#1is-4

ThegroupdelayofFilter#2is–4.5

ThegroupdelayofFilter#3is-4

ThegroupdelayofFilter#4is–4.5

5.3STABILITYTEST

AccordingtoProgrammeP4_4.

Answers:

Q5.23Thepole-zeroplotsofH1（zandH2（zobtainedusingzplaneareshownbelow:

Fromtheabovepole-zeroplotsweobservethat-

Q5.24UsingProgramP4_4wetestedthestabilityofH1（zandarriveatthefollowingstabilitytestparameters{ki}:

-0.99890.8500

FromtheseparametersweconcludethatH1（zis_____stable________.

UsingProgramP4_4wetestedthestabilityofH2（zandarriveatthefollowingstabilitytestparameters{ki}:

-1.00050.8500

FromtheseparametersweconcludethatH2（zis____unstable_________.5.4spectrograms

Background:

Aspectrogramofatimesignalisatwo-dimensionrepresentationthatdisplaystimeinitshorizontalaxisandfrequencyinitsverticalaxis.Agrayscaleistypicallyusedtoindicatetheenergyateachpoint.“white”:

lowenergy,“black”:

highenergy.

Tasks:

ProduceaspectrograminMatlabandthencompareitwithspecgram.minMatlabto

assurewhatyoudidisright.

Outline:

Toproduceaspectrogram,youwillneedtoassemblemanyindividualspectralslices-producedviatheFouriertransform-fromsuccessivesegmentsofaspeechsignal.The

basicoperationsrequiredare

•loadingaspeechsignal

•selectingindividualtimesegmentsofthespeechsignal

•usingtheFFTtotransformfromthetimedomaintothefrequencydomain

•extractingthelogmagnitudespectrum

•assemblingthesequenceofspectraintoa2Dtime-frequencymatrix

•displayingthematrixasanimage

YouwilllearnhowtowriteMATLABscriptsandfunctionstoproducespectrogramsfromsignals.YouwillalsouseMATLAB'splottingfacilitiestodisplaysignalsasameansofcheckingthatyourcodeisdoingtherightthing.

Q5.25

a.LoadthesuppliedspeechsignalintoMATLABusingthecommand:

[y,fs,nbits]=wavread（'oilyrag.wav';

b.Designaprogrammetodividethisspeechintoseveralframes,andcompareitwiththe

command‘buffer’.

x=wavread（'oilyrag.wav';

%调用自己的分帧函数

w=256;

ov=128;

y=fenzhen（x,w,ov;

%调用matlab的分帧函数

z=buffer（x,w,ov;

functiony=fenzhen（x,w,ov

%===分帧函数

if（size（x,1>size（x,2

x=x';

s=length（x;

ifnargin<2

w=256;

end

ifnargin<3

ov=w/2;

end

h=w-ov;

c=1;

ncols=1+fix（（s-w/h;

d=zeros（w,ncols;

forb=0:

h:

（s-w

d（:

c=x（（b+1:

（b+w';

c=c+1;

end

y=d;

c.Designaprogrammetosynthesizetheframedspeechtoasectionofspeech.functiony=hecheng（x,ov

%==合成语音

[framlengframnum]=size（x;

ifnargin<2

ov=framleng/2;

end

h=framleng-ov;

sigleng=h*（framnum-1+framleng;

d=zeros（1,sigleng;

fori=1:

framnum

d（（（i-1*h+1:

（（i-1*h+framleng=d（（（i-1*h+1:

（（i-1*h+framleng+x（:

i';

y=d;

d.Designaprogrammetorepresentthespectrogramofasectionofspeech,andcompareitwiththecomman

d‘specgram’.

[x,sr]=wavread（'Tea_10kHz_16bit.wav';

%调用自己的语谱程序

y=my_specgram（x,sr,256,128;

%调用matlab的语谱程序

figure,

specgram（xfunctiony=my_specgram（x,sr,w,ov%====语谱程序if（size（x,1>size（x,2x=x';ends=length（x;ifnargin<2sr=1;endifnargin<3w=256;endifnargin<4ov=w/2;endh=w-ov;win=hanning（w';c=1;ncols=1+fix（（s-w/h;d=zeros（1+w/2,ncols;forb=0:

h:

（s-wu=win.*x（（b+1:

（b+w;t=fft（u;%ƵÓòÊý¾Ýd（:

c=t（1:

（1+w/2';c=c+1;endtt=[0:

h:

（s-w]/sr;ff=[0:

（w/2]*sr/w;ifnargout<1axisxyxlabel（'Time/s';ylabel（'Frequency/Hz';elsey=d;imagesc（tt,ff,20*log10（abs（d;axisxyxlabel（'Time/s';ylabel（'Frequency/Hz'endimagesc（tt,ff,20*log10（abs（d;Date:

Signature: