Digital Audio Effects

Having learned to make basic sounds from basic waveforms and more advanced synthesis methods lets see how we can add some digital audio effects. These may be applied:
212

As part of the audio creation/synthesis stage to be subsequently

ltered, (re)synthesised

At the end of the audio chain as part of the production/mastering

phase.

Effects can be applied in different orders and sometimes in a

parallel audio chain.

The order of applying the same effects can have drastic differences
in the output audio.

Selection of effects and the ordering is a matter for the sound you
wish to create. There is no absolute rule for the ordering.
Back Close

Effect Types Parameters Typical Guitar (and and other) Effects Pipeline
Some ordering is standard for some audio processing, E.g : all effect illustration below. You can use modules or selectively set certain Linking Effects Compression Distortion EQ Noisetogether Redux Amp Sim modules to on or off. Modulation Delay Reverb
The patches of the G1/G1X consist of eight serially linked effect modules, as shown in the
213

Common for guitar effects pedal.

Effect modules COMP/EFX
Compressor Auto Wah Booster Tremolo Phaser

DRIVE
FD Clean VX Clean HW Clean US Blues BG Crunch

EQ

ZNR
ZNR

AMP
AMP Sim.

MODULATION
Chorus Ensemble Flanger Step Pitch Shift

DELAY
Delay Tape Echo Analog Delay Ping Pong Delay

REVERB
Hall Room Spring Arena Tiled Room

Effect types

* Manufacturer names and product names mentioned in this listing are trademarks or registered trademarks of their respective owners. The names are used only to illustrate sonic characteristics and do not indicate any afliation with ZOOM CORPORATION. For some effect modules, you can select an effect type from several possible choices. For example, the MODULATION module comprises Chorus, Flanger, and other effect types. The REVERB module comprises Hall, Room, and other effect types from which you can choose one.

Back Close

Classifying Effects
Audio effects can be classied by the way do their processing: Basic Filtering Lowpass, Highpass lter etc,, Equaliser: see CM2202 Notes
214

Time Varying Filters Wah-wah, Phaser Delays Vibrato, Flanger, Chorus, Echo Modulators Ring modulation, Tremolo, Vibrato Non-linear Processing Compression, Limiters, Distortion, Exciters/Enhancers Spacial Effects Panning, Reverb (see CM2202 Notes ), Surround Sound Some of the above studied here.3
See these notes for further information. This additional information may be useful for Coursework but is NOT examinable
3

Back Close

Time-varying Filters
Some common effects are realised by simply time varying a lter in a couple of different ways: Wah-wah A bandpass lter with a time varying centre (resonant) frequency and a small bandwidth. Filtered signal mixed with direct signal. Phasing A notch lter, that can be realised as set of cascading IIR lters, again mixed with direct signal.
215

Back Close

Wah-wah Example
The signal ow for a wah-wah is as follows:
direct-mix x(n) Time Varying

wah-mix

y (n)

216

BP

where BP is a time varying frequency bandpass lter.

A phaser is similarly implemented with a notch lter replacing

the bandpass lter.

A variation is the M -fold wah-wah lter where M tap delay

bandpass lters spread over the entire spectrum change their centre frequencies simultaneously. delays and narrow bandwidth lters

A bell effect can be achieved with around a hundred M tap

Back Close

Time Varying Filter Implementation: State Variable Filter

In our audio application of time varying lters we now want independent control over the cut-off frequency and damping factor of a lter. (Borrowed from analog electronics) we can implement a State Variable Filter to solve this problem.
217

One further advantage is that we can simultaneously get lowpass,

bandpass and highpass lter output.

Back Close

The State Variable Filter

yh (n) F1 yb (n) F1 yl (n) x(n)

+ T

+ T

218

1 Q1

T T

where:

x(n) yl (n) yb(n) yh(n)

= = = =

input signal lowpass signal bandpass signal highpass signal

Back Close

The State Variable Filter Algorithm

The algorithm difference equations are given by:
219

yl (n) = F1yb(n) + yl (n 1) yb(n) = F1yh(n) + yb(n 1) yh(n) = x(n) yl (n 1) Q1yb(n 1)

with tuning coefcients F1 and Q1 related to the cut-off frequency, fc, and damping, d:

F1 = 2 sin(fc/fs),

and Q1 = 2d

Back Close

MATLAB Wah-wah Implementation

We simply implement the State Variable Filter with a variable frequency, fc . The code listing is wah wah.m:
% % % % % % % % % % % % % % wah_wah.m state variable band pass
220

BP filter with narrow pass band, Fc oscillates up and down the spectrum Difference equation taken from DAFX chapter 2 Changing this from a BP to a BR/BS (notch instead of a bandpass) converts this effect to a phaser yl(n) = F1*yb(n) + yl(n-1) yb(n) = F1*yh(n) + yb(n-1) yh(n) = x(n) - yl(n-1) - Q1*yb(n-1) vary Fc from 500 to 5000 Hz

infile = acoustic.wav; % read in wav sample [ x, Fs, N ] = wavread(infile);

Back Close

%%%%%%% EFFECT COEFFICIENTS %%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % damping factor % lower the damping factor the smaller the pass band damp = 0.05; % min and max centre cutoff frequency of variable bandpass filter minf=500; maxf=3000; % wah frequency, how many Hz per second are cycled through Fw = 2000; %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

221

% change in centre frequency per sample (Hz) delta = Fw/Fs; % create triangle wave of centre frequency values Fc=minf:delta:maxf; while(length(Fc) < length(x) ) Fc= [ Fc (maxf:-delta:minf) ]; Fc= [ Fc (minf:delta:maxf) ]; end % trim tri wave to size of input Fc = Fc(1:length(x));

Back Close

% difference equation coefficients % must be recalculated each time Fc changes F1 = 2*sin((pi*Fc(1))/Fs); % this dictates size of the pass bands Q1 = 2*damp; yh=zeros(size(x)); yb=zeros(size(x)); yl=zeros(size(x)); % first yh(1) = yb(1) = yl(1) = % create emptly out vectors
222

sample, to avoid referencing of negative signals x(1); F1*yh(1); F1*yb(1);

% apply difference equation to the sample for n=2:length(x), yh(n) = x(n) - yl(n-1) - Q1*yb(n-1); yb(n) = F1*yh(n) + yb(n-1); yl(n) = F1*yb(n) + yl(n-1); F1 = 2*sin((pi*Fc(n))/Fs); end %normalise maxyb = max(abs(yb)); yb = yb/maxyb;

Back Close

% write output wav files wavwrite(yb, Fs, N, out_wah.wav); figure(1) hold on plot(x,r); plot(yb,b); title(Wah-wah and original Signal);

223

Back Close

Wah-wah MATLAB Example (Cont.)

The output from the above code is (red plot is original audio):
Wah!wah and original Signal 1 0.8 0.6 0.4 0.2 0 !0.2 !0.4 !0.6 !0.8 !1 0

224

10

15 x 10
4

Click here to hear: original audio, wah-wah ltered audio.

Back Close

Delay Based Effects

Many useful audio effects can be implemented using a delay structure:

Sounds reected of walls

225

In a cave or large room we here an echo and also reverberation takes place this is a different effect see later If walls are closer together repeated reections can appear as parallel boundaries and we hear a modication of sound colour instead.

Vibrato, Flanging, Chorus and Echo are examples of delay effects

Back Close

Basic Delay Structure

We build basic delay structures out of some very basic FIR and IIR lters:

Combination of FIR and IIR gives the Universal Comb Filter

We use FIR and IIR comb lters

226

Back Close

FIR Comb Filter

This simulates a single delay:

The input signal is delayed by a given time duration, .

The delayed (processed) signal is added to the input signal some amplitude gain, g The difference equation is simply: y (n) = x(n) + gx(n M ) The transfer function is: H (z ) = 1 + gz M
with M = /fs

227

Back Close

FIR Comb Filter Signal Flow Diagram

x(n)

y (n)

228

TM

x(n M )

Back Close

FIR Comb Filter MATLAB Code

rcomb.m:
x=zeros(100,1);x(1)=1; % unit impulse signal of length 100 g=0.5; %Example gain Delayline=zeros(10,1); % memory allocation for length 10 for n=1:length(x); y(n)=x(n)+g*Delayline(10); Delayline=[x(n);Delayline(1:10-1)]; end;

229

Back Close

IIR Comb Filter

This simulates a single delay:

Simulates endless reections at both ends of cylinder. We get an endless series of responses, y (n) to input, x(n). The input signal circulates in delay line (delay time ) that is fed
back to the input..

230

Each time it is fed back it is attenuated by g . Input sometime scaled by c to compensate for high amplication
of the structure.

The difference equation is simply: The transfer function is:

y (n) = Cx(n) + gy (n M ) H (z ) =

with M = /fs

c 1 gz M

Back Close

IIR Comb Filter Signal Flow Diagram

x(n)

y (n)

231

TM

y (n M )

Back Close

IIR Comb Filter MATLAB Code

iircomb.m:
x=zeros(100,1);x(1)=1; % unit impulse signal of length 100 g=0.5; Delayline=zeros(10,1); % memory allocation for length 10 for n=1:length(x); y(n)=x(n)+g*Delayline(10); Delayline=[y(n);Delayline(1:10-1)]; end;

232

Back Close

Universal Comb Filter

The combination of the FIR and IIR comb lters yields the Universal Comb Filter:

Basically this is an allpass lter with an M sample delay operator

and an additional multiplier, FF.
BL

233

x(n)

TM

x(n M )

FF

y (n)

FB

Parameters: FF = feedforward, FB = feedbackward, BL = blend

Back Close

Universal Comb Filter Parameters

234

Universal in that we can form any comb lter, an allpass or a delay: BL FB FF FIR Comb 1 0 g IIR Comb 1 g 0 Allpass a a 1 delay 0 0 1

Back Close

Universal Comb Filter MATLAB Code

unicomb.m:
235

x=zeros(100,1);x(1)=1; % unit impulse signal of length 100 BL=0.5; FB=-0.5; FF=1; M=10; Delayline=zeros(M,1); % memory allocation for length 10 for n=1:length(x); xh=x(n)+FB*Delayline(M); y(n)=FF*Delayline(M)+BL*xh; Delayline=[xh;Delayline(1:M-1)]; end;

Back Close

Vibrato - A Simple Delay Based Effect

Vibrato Varying the time delay periodically
(cf. Doppler effect) we here a change in pitch. to vary the delay.

If we vary the distance between and observer and a sound source Implementation: A Delay line and a low frequency oscillator (LFO) Only listen to the delay no forward or backward feed. Typical delay time = 510 Ms and LFO rate 514Hz.

236

Back Close

Vibrato MATLAB Code

vibrato.m function, Use vibrato eg.m to call function:
function y=vibrato(x,SAMPLERATE,Modfreq,Width) ya_alt=0; Delay=Width; % basic delay of input sample in sec DELAY=round(Delay*SAMPLERATE); % basic delay in # samples WIDTH=round(Width*SAMPLERATE); % modulation width in # samples if WIDTH>DELAY error(delay greater than basic delay !!!); return; end; MODFREQ=Modfreq/SAMPLERATE; % modulation frequency in # samples LEN=length(x); % # of samples in WAV-file L=2+DELAY+WIDTH*2; % length of the entire delay Delayline=zeros(L,1); % memory allocation for delay y=zeros(size(x)); % memory allocation for output vector
237

Back Close

for n=1:(LEN-1) M=MODFREQ; MOD=sin(M*2*pi*n); ZEIGER=1+DELAY+WIDTH*MOD; i=floor(ZEIGER); frac=ZEIGER-i; Delayline=[x(n);Delayline(1:L-1)]; %---Linear Interpolation----------------------------y(n,1)=Delayline(i+1)*frac+Delayline(i)*(1-frac); %---Allpass Interpolation-----------------------------%y(n,1)=(Delayline(i+1)+(1-frac)*Delayline(i)-(1-frac)*ya_alt); %ya_alt=ya(n,1); end

238

Back Close

Vibrato MATLAB Example (Cont.)

The output from the above code is (red plot is original audio):
0.3 Vibrato First 500 Samples

0.2

239
0.1 0

!0.1

!0.2

!0.3

!0.4

50

100

150

200

250

300

350

400

450

500

Click here to hear: original audio, vibrato audio.

Back Close

Comb Filter Delay Effects: Flanger, Chorus, Slapback, Echo

A few popular effects can be made with a comb lter (FIR or IIR)
and some modulation sound outputs:
240

Flanger, Chorus, Slapback, Echo same basic approach but different

Effect Delay Range (ms) Modulation Resonator 0 . . . 20 None Flanger 0 . . . 15 Sinusoidal ( 1 Hz) Chorus 10 . . . 25 Random Slapback 25 . . . 50 None Echo > 50 None

Slapback (or doubling) quick repetition of the sound,

Flanging continuously varying LFO of delay, Chorus multiple copies of sound delayed by small random delays

Back Close

Flanger MATLAB Code

anger.m:
% Creates a single FIR delay with the delay time oscillating from % Either 0-3 ms or 0-15 ms at 0.1 - 5 Hz infile=acoustic.wav; outfile=out_flanger.wav; % read the sample waveform [x,Fs,bits] = wavread(infile); % parameters to vary the effect % max_time_delay=0.003; % 3ms max delay in seconds rate=1; %rate of flange in Hz index=1:length(x); % sin reference to create oscillating delay sin_ref = (sin(2*pi*index*(rate/Fs))); %convert delay in ms to max delay in samples max_samp_delay=round(max_time_delay*Fs); % create empty out vector y = zeros(length(x),1);
Back Close 241

% to avoid referencing of negative samples y(1:max_samp_delay)=x(1:max_samp_delay); % set amp suggested coefficient from page 71 DAFX amp=0.7; % for each sample for i = (max_samp_delay+1):length(x), cur_sin=abs(sin_ref(i)); %abs of current sin val 0-1 % generate delay from 1-max_samp_delay and ensure whole number cur_delay=ceil(cur_sin*max_samp_delay); % add delayed sample y(i) = (amp*x(i)) + amp*(x(i-cur_delay)); end % write output wavwrite(y,Fs,outfile);
242

Back Close

Flanger MATLAB Example (Cont.)

The output from the above code is (red plot is original audio):
Flanger and original Signal 1.5

243
0.5

!0.5

!1

!1.5 0

10

15 x 10
4

Click here to hear: original audio, anged audio.

Back Close

Modulation
Modulation is the process where parameters of a sinusoidal signal (amplitude, frequency and phase) are modied or varied by an audio signal. We have met some example effects s that could be considered as a class of modulation already: Amplitude Modulation Wah-wah, Phaser Frequency Modulation Audio synthesis technique (Already Studied) Phase Modulation Vibrato, Chorus, Flanger We will now introduce some Modulation effects.
244

Back Close

Ring Modulation
Ring modulation (RM) is where the audio modulator signal, x(n) is multiplied by a sine wave, m(n), with a carrier frequency, fc .

This is very simple to implement digitally: y (n) = x(n).m(n) Although audible result is easy to comprehend for simple signals

245

things get more complicated for signals having numerous partials

If the modulator is also a sine wave with frequency, fx then one hears the sum and difference frequencies: fc + fx and fc fx , for
example.

When the input is periodic with at a fundamental frequency, f0, then a spectrum with amplitude lines at frequencies |kf0 fc | Used to create robotic speech effects on old sci- movies and can
create some odd almost non-musical effects if not used with care. (Original speech )
Back Close

MATLAB Ring Modulation

Two examples, a sine wave and an audio sample being modulated by a sine wave, ring mod.m
filename=acoustic.wav; % read the sample waveform [x,Fs,bits] = wavread(filename); index = 1:length(x); % Ring Modulate with a sine wave frequency Fc Fc = 440; carrier= sin(2*pi*index*(Fc/Fs)); % Do Ring Modulation y = x.*carrier; % write output wavwrite(y,Fs,bits,out_ringmod.wav);
246

Click here to hear: original audio, ring modulated audio.

Back Close

MATLAB Ring Modulation: Two sine waves

% Ring Modulate with a sine wave frequency Fc Fc = 440; carrier= sin(2*pi*index*(Fc/Fs)); %create a modulator sine wave frequency Fx Fx = 200; modulator = sin(2*pi*index*(Fx/Fs)); % Ring Modulate with sine wave, freq. Fc y = modulator.*carrier; % write output wavwrite(y,Fs,bits,twosine_ringmod.wav);

247

Output of Two sine wave ring modulation (fc = 440, fx = 380)

1 0.8 0.6 0.4 0.2 0 !0.2 !0.4 !0.6 !0.8 !1 0

50

100

150

200

250

Click here to hear: Two RM sine waves (fc = 440, fx = 200)

Back Close

Amplitude Modulation
Amplitude Modulation (AM) is dened by:

y (n) = (1 + m(n)).x(n) Normalise the peak amplitude of M(n) to 1. is depth of modulation = 1 gives maximum modulation = 0 tuns off modulation

248

When x(n) and m(n) both sine waves with frequencies fc and fx respectively we here three frequencies: carrier, difference and sum: fc , fc fx , fc + fx .
Back Close

m(n) is a low-frequency oscillator modulator.

x(n) is the audio carrier signal

Amplitude Modulation: Tremolo

A common audio application of AM is to produce a tremolo effect:

Set modulation frequency of the sine wave to below 20Hz

The MATLAB code to achieve this is tremolo1.m
% read the sample waveform filename=acoustic.wav; [x,Fs,bits] = wavread(filename); index = 1:length(x); Fc = 5; alpha = 0.5; trem=(1+ alpha*sin(2*pi*index*(Fc/Fs))); y = trem.*x; % write output wavwrite(y,Fs,bits,out_tremolo1.wav);

249

Click here to hear: original audio, AM tremolo audio.

Back Close

Tremolo via Ring Modulation

If you ring modulate with a triangular wave (or try another waveform) you can get tremolo via RM, tremolo2.m
% read the sample waveform filename=acoustic.wav; [x,Fs,bits] = wavread(filename); % create triangular wave LFO delta=5e-4; minf=-0.5; maxf=0.5; trem=minf:delta:maxf; while(length(trem) < length(x) ) trem=[trem (maxf:-delta:minf)]; trem=[trem (minf:delta:maxf)]; end %trim trem trem = trem(1:length(x)); %Ring mod with triangular, trem y= x.*trem; % write output wavwrite(y,Fs,bits,out_tremolo2.wav);

250

Click here to hear: original audio, RM tremolo audio.

Back Close

Non-linear Processing
Non-linear Processors are characterised by the fact that they create (intentional or unintentional) harmonic and inharmonic frequency components not present in the original signal. Three major categories of non-linear processing: Dynamic Processing: control of signal envelop aim to minimise harmonic distortion Examples: Compressors, Limiters Intentional non-linear harmonic processing: Aim to introduce strong harmonic distortion. Examples: Many electric guitar effects such as distortion Exciters/Enhancers: add additional harmonics for subtle sound improvement.
Back Close 251

Overdrive, Distortion and Fuzz

Distortion plays an important part in electric guitar music, especially rock music and its variants. Distortion can be applied as an effect to other instruments including vocals. Overdrive Audio at a low input level is driven by higher input levels in a non-linear curve characteristic Distortion a wider tonal area than overdrive operating at a higher non-linear region of a curve Fuzz complete non-linear behaviour, harder/harsher than distortion
252

Back Close

Overdrive
For overdrive, Symmetrical soft clipping of input values has to be performed. A simple three layer non-linear soft saturation scheme may be: for 0 x < 1/3 2x 2 3x) f (x) = 3(2 for 1/3 x < 2/3 3 1 for 2/3 x 1
253

In the lower third the output is liner multiplied by 2.

response

In the middle third there is a non-linear (quadratic) output Above 2/3 the output is set to 1.

Back Close

MATLAB Overdrive Example

The MATLAB code to perform symmetrical soft clipping is, symclip.m:
function y=symclip(x) % y=symclip(x) % "Overdrive" simulation with symmetrical clipping % x - input N=length(x); y=zeros(1,N); % Preallocate y th=1/3; % threshold for symmetrical soft clipping % by Schetzen Formula for i=1:1:N, if abs(x(i))< th, y(i)=2*x(i);end; if abs(x(i))>=th, if x(i)> 0, y(i)=(3-(2-x(i)*3).2)/3; end; if x(i)< 0, y(i)=-(3-(2-abs(x(i))*3).2)/3; end; end; if abs(x(i))>2*th, if x(i)> 0, y(i)=1;end; if x(i)< 0, y(i)=-1;end; end; end;

254

Back Close

MATLAB Overdrive Example (Cont.)

An overdriven signal looks like this , overdrive eg.m:
Overdriven Signal

255

% read the sample waveform filename=acoustic.wav; [x,Fs,bits] = wavread(filename); % call symmetrical soft clipping % function y = symclip(x); % write output wavwrite(y,Fs,bits,... out_overdrive.wav); figure(1); hold on plot(y,r); plot(x,b); title(Overdriven Signal);

1 0.8 0.6 0.4 0.2 0 !0.2 !0.4 !0.6 !0.8 !1 0

10

15 x 10
4

Click here to hear: original audio, overdriven audio.

Back Close

Distortion/Fuzz
A non-linear function commonly used to simulate distortion/fuzz is given by:

This a non-linear exponential function:

x 2 (1 ex /|x|) f ( x) = |x|

256

The gain, , controls level of distortion/fuzz.

for output.

Common to mix part of the distorted signal with original signal

Back Close

MATLAB Fuzz Example

The MATLAB code to perform symmetrical soft clipping is, fuzzexp.m:
257

function y=fuzzexp(x, gain, mix) % y=fuzzexp(x, gain, mix) % Distortion based on an exponential function % x - input % gain - amount of distortion, >0-> % mix - mix of original and distorted sound, 1=only distorted q=x*gain/max(abs(x)); z=sign(-q).*(1-exp(sign(-q).*q)); y=mix*z*max(abs(x))/max(abs(z))+(1-mix)*x; y=y*max(abs(x))/max(abs(y));

Note: function allows to mix input and fuzz signals at output

Back Close

MATLAB Fuzz Example (Cont.)

An fuzzed up signal looks like this , fuzz eg.m:

258
Fuzz Signal 1 0.8

filename=acoustic.wav; % read the sample waveform [x,Fs,bits] = wavread(filename); % Call fuzzexp gain = 11; % Spinal Tap it mix = 1; % Hear only fuzz y = fuzzexp(x,gain,mix); % write output wavwrite(y,Fs,bits,out_fuzz.wav);

0.6 0.4 0.2 0 !0.2 !0.4 !0.6 !0.8 !1 0

10

15 x 10
4

Click here to hear: original audio, Fuzz audio.

Back Close