Lecture Notes 09PC602 DIGITAL SIGNAL PROCESSING VI ...

219
Lecture Notes 09PC602 DIGITAL SIGNAL PROCESSING VI Sem B.E (I.T) Dr. J. SASIKALA Associate Professor Department of Information Technology

Transcript of Lecture Notes 09PC602 DIGITAL SIGNAL PROCESSING VI ...

Lecture Notes

09PC602 DIGITAL SIGNAL PROCESSING

VI Sem B.E (I.T)

Dr. J. SASIKALA Associate Professor

Department of Information Technology

1  

Dr. J. Sasikala Associate Professor

IIR Filters Unit-3

   

 

2  

 

 

3  

 

 

 

 

 

4  

 

5  

 

6  

 

7  

 

8  

 

9  

 

10  

 

11  

 

12  

 

13  

 

14  

 

15  

 

16  

 

17  

 

18  

 

19  

 

20  

 

21  

 

22  

 

23  

 

24  

 

25  

 

26  

 

27  

 

28  

 

29  

 

30  

 

31  

 

32  

 

33  

 

34  

 

35  

 

36  

 

37  

 

38  

 

39  

 

40  

 

41  

 

42  

 

43  

 

44  

 

45  

 

46  

 

47  

 

48  

 

49  

 

50  

 

51  

 

52  

 

53  

 

54  

 

55  

 

56  

 

57  

 

58  

 

59  

 

60  

 

61  

 

62  

 

63  

 

64  

 

65  

 

66  

 

67  

 

68  

 

69  

 

70  

 

71  

 

72  

 

73  

 

74  

 

75  

 

76  

 

77  

 

 

78  

 

79  

 

80  

 

81  

 

82  

 

83  

 

84  

 

85  

 

 

 

 

 

 

 

 

 

 

 

 

Unit-4 - Finite Word Length Registers

UNIT‐V  Multirate Digital Signal Processing 

 

 

 

 

 

 

F

 

ig. Spectra of the inteermediate aand decimaated sequen

 

nce 

 

 

 

 

 

Fig. Spectra of the Original, intermediate and other sequences 

 

 

 

 

 

 

 

Fig. An illustration of interpolation by a rational factor (L=3, M=2) 

 

 

 

 

 

 

 

 

F

fi

   

ig.  Polypha

ilter 

   

ase decompposition of 

      

(a) a decimmation filter   (b) an in

 

nterpolation

 

 

 

 

 

 

 

 

 

 

 

 

 

Fig. Digital to Analogue conversion for a CD player using x8 oversampling 

 

 

 

   

 

 

Fig. Simplified block diagram of single‐bit ADC scheme 

 

Speech Compression 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   

 

Filter into two bands: 50 Hz‐3.5 KHz: sample at 8 kHz, encode at 48 KB/s 3.5 KHz‐ 7 KHz: sample at 16 kHz, encode at 16 KB/s  

 

 

 

 

4 KHz‐7 KHz 

 

 

 

 

                                 Vocal tract excitation parameters (LPC

Coefficients)

Adaptive Filters 

Adaptive Filters ‐ Applications 

New Trends in Adaptive Filtering

Musical Sound Processing The audio effects are artificially generated using various signal processing circuits and devices, and increasingly by digital signal processing techniques, often referred as Musical Sound processing. All musical programs are produced in basically two stages:

Time Domain Operation Commonly used time-domain operations are:

   

Sound from each individual instrument is recorded in an acoustically inert studio on a single track of a multi-track

tape recorder

The signals from each track are manipulated by the sound engineer to add special audio effects and are combined in a mix-down system to

finally generate the stereo recording on a two-track tape

Echo armagnitu For exalater cacharact

To genethe exp One ca

 

re simply ude respo

ample, then be simperized by

erate a fixonentially

n use an

generatedonse, such

e direct soply genera

the differ

xed numbey decaying

FIR filter w

Singl

d by delayh a filter is

ound and aated by thrence equ

Multip

er of multg amplitud

with a tran

H(z) =

 

le Echo

y units. Bes known a

a single ee FIR filte

uation:

ple Echo

tiple echoedes.

nsfer func

o Filter

ecause ofas comb fi

echo appeer shown i

o Filter

es spaced

ction of th

f the com lter.

earing R sin Fig., wh

d R samp

e form:

like shap

sampling phich is

pling perio

e of the

periods

ods with

The

The for tw

A nuOne

It wacombdelay

One the oflang

Flang

The musismal

The

reverbera

IIR combwo reason

Its magn"coloratilistening

The outpgenerateobservesound.

umber of such effe

as createdbining they.

way of vaoperators ging.

ging Effec

chorus efical piecell timing d

phasing

ation is co

b filter by ns, which

nitude resion" of mg.

put echo ded by a un

ed in a rea

special soect is flang

d by feedieir delaye

arying timthumb on

ct:

Chor

ffect is ace at the saifferences

effect

Rev

onsidered

itself doeare:

sponse is many mu

density ginit impuls

al room th

F

ound effeging.

ng the saed outputs

me is to slon the flan

rus Gen

chieved wame time s between

is produc

verbera

o densely

es not pro

not constusical sou

ven by nuse at the inus causin

Flangin

ects are o

ame musics while v

ow down nge of the

nerator

when sevebut with

n their sou

ced by

ation

y packed

ovide natu

tant for alund that

umber of enput is mung "flutterin

g

often used

cal piece varying th

one of thee feed ree

and Ph

eral musicsmall cha

unds.

processin

echoes.

ural-sound

l frequencare ofte

echoes peuch lower ng" of the

d in the m

to tape ree differen

e tape recel, which i

asing

cians are anges in t

ng the s

ding reve

cies, resuen unplea

er secondthan that

e composi

mix-down

ecorders nce betwe

corders bys led to t

playing tthe amplit

signal th

rberation

ulting in a asant for

d t te

process.

and then een their

y placing he name

he same tude and

rough a

narroscale

Chor

Phas

Thesequamid-fthe bthis r

Adva

Disa

 

owband ned portion

rus Effect

sing Effec

se effectsalizer, thefrequencybass-treblrange.

antages: an informperfect reGuaranteStored a

dvantageSpeed aDSP des(softwareFinite woeconomiperforma

notch filten of the no

:

ct:

Freq

s are ace purposey compone relation

mation caneproducibieed accuralmost inde

s: nd cost- besigns can e etc.) ord-length c conside

ance.

er with votch filter o

quency-

chieved b of equal

nents in thnships by

be conveility-identicacy is onlyefinitely wit

e expensivbe time

problems erations,

 

variable boutput the

-Domain

by passinlizer is tohe range providing

yed, displacal performy determinethout loss

ve with largconsuming

- if only aserious

botch chae original

n Opera

ng the or provide of 1.5 GH boost or

ayed or mamance fromed by the nof informa

ge bandwig plus ne

a limited ndegradatio

aracteristicsignal.

ations

riginal sig"presence

Hz to 3 Gcut to co

anipulatedm unit to unnumber of tion

idth signalseed the ne

number of on may

cs and a

gnals throe" by pea

GHz and tomponnets

d nit. bits used.

s ecessary r

bits is useresult in

adding a

ough an aking the to modify s outside

.

resources

ed due to n system

Imp

Filter

ulse Respo

Ou

r Structu

onse with

utput froSingl

re

Multip

A=0.8 for RRev

Bl

om the le Echo

ple Echo

R=4 verbera

ock Diagr

Algorito Filter

ImpuA=

o Filter

Magniation

ram

hm

lse Respo=0.8, N=6

itude Resp

onse with& R=4

ponse for R

R=7

Imp

Im

pulse Resp

mage E

ponse with

Enhan

h A=0.8 for

ncemeOriginal

Compre

r R=4

ent l Image

essed Ima

ge

   

 

E

 

Imag

Example

ge Enhanc

es

cement

Spatial Domain Enhancement

Enhancement Methods

Noisse Redu

uction

Intensity Adjustment

 

Histog

 

ram Equ

ualizationn

Image Thresholding

   

Gray Level Slicing

Image Rotation

   

Examp

ples:

Conversion M

Methods

Resources Required

   

Examples Original Image Image with Salt & Pepper Noise

Filtered Image Histogram Equalization

Contrast Stretched Image Thresholded Image

   

 

 

Grayscale slicing background

Grayscale slicing without background

180 degree rotation

   

APPLICATIONS Biology Astronomy

Medicines Security, Biometrics

Satellite Imaging Personal Imagery

APPLICATIONS

Example : Audio Sample Rate Conversion

Example : Oversampling ADC