1

EE 5410 Signal Processing

MATLAB Exercise 1

Telephone Touch-Tone Signal Encoding and Decoding

Intended Learning Outcomes:

On completion of this MATLAB laboratory exercise, you should be able to

Generate and decode telephone touch-tone signals
Understand the impact of additive noise in decoding touch-tone signals

Deliverable:

Each student is required to submit a hardcopy answer sheet which contains
only answers to the questions in this manual on or before 23 September 2013.

Background:

Telephone touch-tone pads generate dual tone multiple frequency (DTMF) signals to
dial a telephone. When any key is pressed, the sinusoids of the corresponding row
and column frequencies, which are depicted in Table 1, are generated and summed
to give dual tone. As an example pressing the 5 key generates a signal containing
the sum of the two tones at 770 Hz and 1336 Hz together, and mathematically, it can
be generated as
) 1336 2 cos( ) 770 2 cos( ) ( t t t x t + t =

In fact, the frequencies in Table 1 are chosen to avoid harmonics. No frequency is an
integral multiple of another, the difference between any two frequencies does not
equal any of the frequencies, and the sum of any two frequencies does not equal
any of the frequencies. This makes it easier to detect exactly which tones are
present in the dialled signal in the presence of non-linear line distortion.

Frequencies (Hz) 1209 1336 1477
697 1 2 3
770 4 5 6
852 7 8 9
941 * 0 #
Table 1: DTMF encoding for touch-tone dialling

Decoding of DTMF signals can be achieved via using a simple finite impulse
response (FIR) filter bank which is shown in Figure 1. The filter bank consists of 7
band-pass filters (BPFs) where each filter passes only one of the 7 possible DTMF
frequencies.
2
x[n]
Filter centered
at 770 Hz
y2[n]
Filter centered
at 690 Hz
y1[n]
Filter centered
at 1477 Hz
y7[n]
Detect
dual tones
Detected
key

Figure 1: Block diagram for DTMF signal decoding

When the input ] [n x to the filter bank is a DTMF signal, the outputs from two of the
BPFs should be larger than the rest. If we detect the two largest outputs, the two
corresponding frequencies can be found. These frequencies are then used as row
and column pointers to determine the key from the DTMF code. A good measure of
the output levels can be the peak value at the filter outputs, because when the BPF
is working properly it should pass only one sinusoidal signal and the peak value
would be the amplitude of the sinusoid passed by the filter.

Procedure:

1. If you are not familiar with MATLAB, you can view the MATLAB introduction by
typing intro at the MATLAB prompt. This short introduction will demonstrate
some of the basics of using MATLAB. Or you can explore the MATLAB help
capability which is available at the command line, such as help, help plot
and help clear, where plot and clear are command names. Apart from a
number of MATLAB reference books such as [1]-[2] which can be found in City
Universitys library, many on-line MATLAB resources, including [3]-[4], are
available.

2. Create a file named toneA.m with the following MATLAB code:

clear all
Fs=8000;
Ts=1/Fs;
t=[0:Ts:1];
F_A=440; %Frequency of note A is 440 Hz
A=sin(2*pi*F_A*t);
sound(A,Fs);

Type toneA at the command line and then answer the following:

(a) What is the time duration of A?
(b) How many samples are there in A?
(c) Modify toneA.m by changing Fs=8000 to Fs=800. Can you hear any
difference?
(d) The frequencies of notes B, C#, D, E and F# are 493.88 Hz, 554.37 Hz,
587.33 Hz, 659.26 Hz and 739.99 Hz, respectively. Write a MATLAB file
3
named song.m to produce a piece of music with notes in the following
order : A, A, E, E, F#, F#, E, E, D, D, C#, C#, B, B, A, A. Assign the
duration of each note as 0.3s.

3. Create a file named tone.m with the following MATLAB code:

function x = tone(frequency, observation_length);
% x=tone(frequency, observation_length) is used to generate
% a sinusoidal signal x with frequency and observation
% length specified in the arguments.
fs = 4000;
Ts = 1/fs;
t = [0:Ts:observation_length];
x = cos(2*pi*frequency*t);

Note that tone is a user-defined MATLAB function. Try the following commands:
help tone and y=tone(100,0.01). What are the usages of these two
commands?

4. Write a MATLAB function named dtmfdial.m, to implement a DTMF dialer
based on the frequency table in Table 1. A skeleton of dtmfdial.m is given as
follows:

function xx=dtmfdial(keyName)
%DTMFDIAL Create a DTMF tone
%usage: xx=dtmfdial(keyName)
% keyName = character which is one of the valid key names
% xx = signal vector that corresponds to the DTMF
dtmf.keys = ['1','2','3';
'4','5','6';
'7','8','9';
'*','0','#'];
ff_cols = [1209,1336,1477];
ff_rows = [697;770;852;941];
dtmf.colTones = ones(4,1)*ff_cols;
dtmf.rowTones = ff_rows*ones(1,3);

Complete dtmfdial.m so that it implements the following:

(i) The input to the function is one of the valid key names.

(ii) The output should be a vector of samples at sampling frequency 8000 =
s
f
Hz containing the DTMF tone. Each DTMF signal is the sum of a pair of
unity amplitude sinusoidal signals and the time duration is 0.2s.

(iii) The frequency information is given in two 3 4 matrices, namely,
dtmf.colTones and dtmf.rowTones. To translate a key into the
correct locations of the two matrices, the find function can be used. An
example of using find is:

4
[ii,jj] = find(3==dtmf.keys)

(iv) Play the sound of the DTMF tone using soundsc.

5. One simple way to implement a band-pass FIR filter for DTMF signal decoding is
to use the following impulse response:
L n n
L
n h < s e = 0 ), cos(
1
] [
where e is the center frequency of the band-pass filter and L is the filter length.
Use MATLAB to generate a band-pass filter with t = e 2 . 0 .

(a) Try the cases of 50 = L and 500 = L . Plot the magnitudes of the frequency
spectra of the two filters using freqz. What do you expect about the
magnitude spectrum when L ? An example of using freqz is:

[a,b] = freqz(h); %h is the impulse response
plot(b,abs(a));

(b) Compute the energies of ] [n h for 50 = L and 500 = L . The energy of ] [n h
is defined as
2
1
0
] [n h E
L
n
h
=

=

(c) Which filter will give a better DTMF decoding performance, ] [n h with
50 = L or 500 = L ? Explain your answer.

Note that in general, the impulse response for this simple band-pass FIR filter is

L n
f
n f
L
n h
s
b
< s
|
|
.
|

\
| t
= 0 ,
2
cos
1
] [

where
b
f is the center frequency of the filter and
s
f is the sampling frequency,
both in Hz.

6. Write a MATLAB function named dtmfdetect.m, to implement a DTMF
encoder and decoder in a noisy environment. The requirements of the
dtmfdetect function are given as follows:

(i) The input to the function consists of one of the valid key names, filter
length of the band-pass filters and noise power. That is,
dtmfdetect(1,50,1) will generate a DTMF tone 1 with 50 = L and
the tone is corrupted by a zero-mean white Gaussian noise with power of
1. The output will show the result of the detection, namely, displaying a
message of The detected key is 1.

(ii) Each DTMF signal is the sum of a pair of unity amplitude sinusoidal
signals and the time duration is 0.2s with sampling frequency 8000 =
s
f .

(iii) To add a zero-mean white Gaussian noise to the noise-free DTMF tone,
you can use the randn command. An example of using randn is:
5
noise = sqrt(0.1)*randn(1,10);

where a zero-mean Gaussian noise sequence of length 10 with power of
1 0
2
. = o will be generated.

(iv) To detect the DTMF tone frequencies, you first need to pass the signal to
a filter bank of 7 band-pass filters whose center frequencies are 697 Hz,
770 Hz, 852 Hz, 941 Hz, 1209 Hz, 1336 Hz and 1477 Hz. The DTMF tone
can then be deduced from the two outputs with the largest energy. An
example of producing the output signal given the input and FIR filter
coefficients is

y=conv(x,h); % x is the input and h is the filter
% impulse response

An example of computing the energy of a signal is

energy = sum(y.*y);

Try your dtmfdetect function with various keys, different L ( 50 = L and 500 = L )
and noise powers ( 0
2
= o , 1
2
= o and 50
2
= o ). For each condition, perform 10 trials
and record the number of correct detection in the following table. For example, if the
DTMF tone is 1 and we detect 1 10 times at 50 = L and 0
2
= o , we write 10 in
the corresponding entry.

Key 50 = L
0
2
= o
500 = L
0
2
= o
50 = L
1
2
= o
500 = L
1
2
= o
50 = L
50
2
= o
500 = L
50
2
= o
1
2
3
4
5
6
7
8
9
0
*
#

References:
[1] E.Part-Enander, A.Sjoberg, B.Melin and P.Isaksson, The MATLAB Handbook,
Addison-Wesley, 1996
[2] D.Hanselman and B.Littlefield, Mastering MATLAB 7, Pearson Education
International, 2005
[3] http://www-h.eng.cam.ac.uk/help/tpl/programs/matlab.html
[4] www.mathworks.com/access/helpdesk/help/techdoc/learn_matlab/bqr_2pl.html
[5] L.Schenker, "Pushbutton Calling with a Two-Group Voice-Frequency Code,"
The Bell System Technical Journal, vol.39, no.1, pp.235255, 1960
[6] J.H.McClellan, R.W.Schafer and M.A.Yoder, Signal Processing First, Prentice-
Hall, 2003