fractional delay filters presentation

Upload: akshay

Post on 01-Jun-2018

223 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/9/2019 Fractional Delay Filters Presentation

    1/19

    Fractional Delay Filte

    Aksh

  • 8/9/2019 Fractional Delay Filters Presentation

    2/19

    Introduction

    What do we look for in a fractional delay filter :

    Flat magnitude response.

    Continuous control of the delay.

    Efficient such that fast coefficient update takes place.

    Some applications :

    Synchronization in digital modems

    Speech coding

    Tuning of music

  • 8/9/2019 Fractional Delay Filters Presentation

    3/19

    Ideal Fractional Delay

    )()( txty cc

    A delay element, whose purpose is to delay an incoming continuo

    signalxc(t) by (in seconds).

    j

    c

    id

    c

    j

    eX

    H

    Xe

    )(

    )(Y)(

    )()(Y

    c

    c

    The transfer function Hid() of the delay element can be expresse

    the below equation.

  • 8/9/2019 Fractional Delay Filters Presentation

    4/19

    Discrete-Time System

    Consider a discrete-time delay system where Dis the desire

    y(n) = x(n - D) The spectrum of a discrete-time signal can be expressed by

    the discrete-time Fourier transform (DTFT).

    where D = Dint+ d and 0

  • 8/9/2019 Fractional Delay Filters Presentation

    5/19

    Characteristics of the Ideal FractionalDelay Element

    D

    D

    DeH

    eH

    eH

    idg

    idp

    j

    id

    j

    id

    Djid

    )(

    )(

    )}(arg{

    1|)(|

    ||,)(

    ,

    ,

    Phase delay

    Group delay

    Ideal phase

    response

    The delay element is a linear-phase all pass system.

  • 8/9/2019 Fractional Delay Filters Presentation

    6/19

  • 8/9/2019 Fractional Delay Filters Presentation

    7/19

    Inferences from ideal impulse response

    The impulse response hid(n) of the delay element is a shifte

    sampled version of the sinc function which is infinitely long.

    The filter is not BIBO stable.

    This kind of ideal filter is nonrealizable.

    Some finite-length approximation for the sinc function has t

  • 8/9/2019 Fractional Delay Filters Presentation

    8/19

    Design of fractional delay filters

    Lagrange interpolation design

    Make the error function maximally flat at a certain frequency by sderivative to 0.

    This leads to the Lagrange interpolation formula

    for n= 0,1,2N

    N

    k kn

    kDnh

    0

    )(

    Advantages

    This method gives excellent approximation at low frequencies

    Easy formula for finding out the coefficients.

    Ideal method where good accuracy at high frequencies is not r

  • 8/9/2019 Fractional Delay Filters Presentation

    9/19

    Lagrange interpolation method

    Odd vs. ev

    filters.

    Tradeoff b

    magnitude

    phase delay

    [1]

  • 8/9/2019 Fractional Delay Filters Presentation

    10/19

    Lagrange interpolation method

    Implementation in matlab :

  • 8/9/2019 Fractional Delay Filters Presentation

    11/19

    Lagrange interpolation method

    Fractional delay in a FIR filter :

  • 8/9/2019 Fractional Delay Filters Presentation

    12/19

    Lagrange interpolation method

    Change in the magnitude response

  • 8/9/2019 Fractional Delay Filters Presentation

    13/19

    Least Squared Integral Error Design

    One of the most attractive method for designing realizable FD filters a

    obtained by the truncation of the ideal impulse response.

    This is for 0

  • 8/9/2019 Fractional Delay Filters Presentation

    14/19

    Windowing Method

    A method basically continuation of the Least squares method.

    A method to reduce the Gibbs phenomenon in FIR filter design is to us

    function.

    Here the ideal impulse response is truncated and shaped which is of l

    N+1 with the help of a window function.

    The obtained impulse response does not impact any least squared me

    it is a modification of the least square method.

  • 8/9/2019 Fractional Delay Filters Presentation

    15/19

    Windowing Method

    The coefficients are given by :

    The windowing method is suitable for real-time systems where the fra

    delay is changed since the coefficients can be updated quickly.

    Also this method is fast and easy.

  • 8/9/2019 Fractional Delay Filters Presentation

    16/19

    Windowing Method in matlab

  • 8/9/2019 Fractional Delay Filters Presentation

    17/19

    Windowing Method in matlab

  • 8/9/2019 Fractional Delay Filters Presentation

    18/19

    Windowing Method in matlab

  • 8/9/2019 Fractional Delay Filters Presentation

    19/19

    REFERENCES

    [1] Laakso, Timo I.; Vlimki, Vesa; Karjalainen, Matti; Laine, Unto K. (

    1996), Splitting the unit delay - tools for fractional delay filter design,

    Processing Magazine.

    INTERPOLATED ALLPASS FRACTIONAL-DELAY FILTERS USING ROOT

    DISPLACEMENT. Huseyin Hachabiboglu, Banu G unel, Ahmet M. Kond

    Principles of fractional delay filters; V Valimaki, TI Laakso.

    Fractional Delay Digital Filters ;Javier Diaz-Carmona and Gordana Jova

    Dolecek Institute ITC Celaya, Institute INAOE Puebla, Mexico.