LAB MANUAL DIGITAL SIGNAL PROCESSING

Instructor: Syed Zeeshan Haider Department of Telecom Engineering GC University Faisalabad

DSP Lab An Introduction The DSP lab has both software and hardware modules. In the software component, students carry out a number of computer experiments written in C or MATLAB, illustrating some of the fundamental concepts and applications of digital signal processing, such as quantization and sampling, block processing by convolution, real-time filtering on a sample-by-sample basis, signal enhancement and noise reduction filters, direct, canonical, and cascade realizations of digital filters, spectral analysis by the DFT and FFT, and the design of FIR and IIR digital filters. The choice and number of computer experiments varies from year to year. The hardware part of the lab illustrates the programming of real-time processing algorithms on a concrete DSP chip, the Analog Devices ADSP-2181 fixed-point DSP. The hardware experiments include aliasing and quantization effects; the circular buffer implementation of delays, FIR, and IIR filters; the canceling of periodic interference with notch filters; wavetable generators; and several digital audio effects, and phasers, plain, allpass, and lowpass reverberators, Modem, DTMF generator, and several multi-tap, multi-delay, and stereo-delay type effects. Students are assumed to be familiar with LAB VIEW and MATLAB programming. Most of the required MATLAB/LAB VIEW routines are supplied and the students put them together in their main programs. Lab Guidelines To be accepted and receive full credit, the lab reports/note books must be prepared according to the following guidelines: Include a 12 page discussion of the purposes and objectives of the experiment. Include a short description of the programming ideas you used and a thorough evaluation of the results you obtained. All supporting graphs must be included in this part. The source code of your programs should be attached in your of note book of experiments, cut and pasted to 8.511 paper. MATLAB has an extensive library of DSP routines. But because the purpose of this introductory lab is to teach the inner workings of DSP algorithms, the use of MATLABs built-in DSP routines, such as conv or filter, is allowed only where asked. Students may, however, write their own versions of these routines using the MATLAB language. Attendance is required in all ADSP-2181 sessions of the lab It is not possible to receive a grade of A if one of these sessions is missed. In addition, a 12 page lab report on each hardware lab must be submitted at the next hardware lab. Experiment reports in notebooks in are due at the beginning of the class period on the due date. Please, note that late labs will not be accepted.

1. 2. 3. 4. 5. 6. 7.

I. LAB-1 MATLAB TUTORIAL

Objective: In this lab you will cover the following basic things: Using Matlab as a numerical calculator Entering row vectors and column vectors Entering matrices and their manipulation, Generating Signals Forming Signal addition, products Doing two dimensional signal manipulation Using Matlab to solve linear System equations, Implementing Digital Systems Matlab functions that operate on arrays Plotting basic graphs using Matlab.

II. LAB 2
Experiment-1: Hardware & Software Setup: Speedy-33 Interfacing:
The National Instruments LabVIEW DSP Module can be easily interfaced with NI SPEEDY-33 and Texas Instruments DSP Starter Kits, including the TI 6713 and TI 6711 to implement and analyze DSP systems. With Lab VIEW, users use graphical programming methods to learn and develop applications and they can take advantage of the LabVIEW DSP Module to quickly implement DSP fundamentals on DSP hardware without going through the tedious task of having to write any C code, assembly, or script source.

1.2 What is Speedy-33?

SPEEDY stands for Signal Processing Educational Engineering Device for Youth. It connects to a standard PC via a Universal Serial Bus (USB) host port. The VC33 DSP Education Kit from NI includes the SPEEDY-33 and the LabVIEW DSP Embedded software.

Figure : Speedy 33 Board Overview

There are several things on this board that you should observe. They are: (1) On the lower left end of the picture is the power port. The input voltage must be 9 Volt DC VDC), at 500 milliamps. (2) Right above it is the connector that attaches to a USB port of your PC. (3) The two small rectangular devices at the top left and top right are onboard microphones. (4) The biggest thing on the board is the processor or DSP. Texas Instruments (TI) makes this particular DSP. TI makes many different DSPs; the TMS320VC33 tells which chip this one is. This DSP chip has 150 written on the top. The processor is running at 150 MHz. (5) Right next to the DSP is the flash memory.

(6) One of the chips is the Audio Interface Controller. Its job is to convert the analog signal from the microphone into a digital form the processor can use. (7 & 8) The audio input level jumpers (J1, J2) control the amount of gain applied to the input audio signals. It is used to choose between Line Input (11) and On-Board Microphones (3). (9)When power is supplied to the device, the power LED lights. (10) The reset button is a small push button that manually resets the DSP and is usually used in the event of a software or hardware freeze. (11, 12) The two connectors on the right end are Audio Stereo Input Port &Audio Stereo Output Port. (13 & 14) There are eight lines of digital Input-Output (I/O) on the board: Input Ports are represented by 8 switches (lower right) and Output Ports are represented by 8 LEDs (right bottom). The digital I/O lines can be programmed with the LabVIEW DSPModule. (15) The Flash Boot Jumper controls whether the DSP will attempt to boot from the flash memory, which is the default setting. It must be enabled to run the device in standalone mode. (16) The simple 20-pin expansion header allows for easy interface to external hardware. The header includes power, ground, eight digital inputs, and eight digital outputs under DSP control. (17) Two connectors make up the standard expansion analog I/O connectors, that can be used for optional daughter modules. (18) The flash write enable jumper controls whether the DSP can write to the flash memory or not. This is useful in write protecting the DSP algorithm for production purposes.

1.5 Hardware & Software Setup

For all the experiments using the SPEEDY-33 board, you will have to do the following instructions at the beginning of each experiment. Log on to the computer Connect a USB cable between the USB Port of the SPEEDY-33 and the USB port of your PC. The SPEEDY-33 is powered by the USB port. Make sure that the Power LED on the SPEEDY33 is on. Connect the speakers to the Audio Stereo Output Port on the SPEEDY-33. The on-board microphones will be used as the input to the DSP board. The next step is to choose a target for the DSP program to be downloaded to. Using the Lab VIEW DSP module, it is very simple to do so as illustrated in figure below By clicking on the Execution Target pull-down menu, one can see that the LabVIEW DSP module supports only three DSP targets: NI Speedy-33, TI C671 1 and TI C6713. No other DSP hardware can be connected to the LabVIEW DSP module. In our case, you will have to select Speedy-33 as the target.To create a new VI, click New and then select Blank VI. As a result, two blank windows will appear which form the basic elements of a Lab VIEW application also called Virtual Instrument (VI). The steps in the Hardware & Software setup will always be done at the beginning of each experiment involving the SPEEDY-33. The experiments instructions will differ in the next sections.

Lab2
Experiment-2:
A/D and D/A Conversion
In this section, you will develop a simple VI. As a starting point, you will read an analog input from the A/D channel and play it back using the D/A channel. Circuit Diagram

Procedure:
1.

Click on the Elemental I/O sub-palette and bring out the Analog Input Elemental I/O node and place it on the block diagram.

2. You can select from a variety of sampling rates (8000->48000 Hz) to configure it, set a sampling rate of 48 kHz and a frame size of 256.
3. Click on the OK button to close the configuration panel for Analog Input and return to

the block diagram (The Analog Input block is equivalent to an A/D converter. The Analog Output block is equivalent to the D/A converter). 4. Repeat the same steps to place and configure an Analog Output Elemental I/O node on the block diagram. 5. Finally, connect the output terminal of Analog Input to the input terminal of the Analog Output. 6. Place a while Loop to the VI by going to the Structures sub-palette and selecting While Loop.
7.

The final VI should look similar to the one shown in figure above.

8. To run the VI, click on the run but ton in the upper left corner of the Block Diagram window. The LabVIEW DSP Module status monitor window opens as shown in figure below. It should show that the program is compiled with 0 errors this it is downloading to the target.

Try speaking through the onboard microphones while listening through the headphones. Can you pick up the sound? Try to get further form the board. Can you still pick up the sound? Stop running the program by clicking the stop button in the Front Panel window as shown in Figure below. To make the microphone more sensitive, you can add a gain to the Analog Input before feeding it to the Analog Output. Items to be checked

Hardwate & Software Setup A/D and D/A Conversion Sound Picking by onboard microphones while listening through the headphones.

III. LAB -3
Objective:
Lab work in this course will comprise of two components i.e. lab exercises & lab tasks. You will have to complete the lab exercises within the stipulated lab timings and show your work/code/waveforms etc. For Lab tasks you will have the whole week to work. Lab tasks will be due before the next weeks lab. Lab Exercise 1 Procedure: The sample code below decomposes a rectangular pulse shaped signal into its even and odd components. Type help command in the command window to see information about a particular command. E.g. help fliplr. n = [-10:10]; %index varies from -10 to 10 x = [zeros(1,10) ones(1,11)]; %x is a rectangular pulse from 0 to 10 f = fliplr(x); % type 'help fliplr' on matlab command window to see what does this do even = (x+f)/2; %make even and.. odd = (x-f)/2; %odd signals from x y = even + odd; %'y' should be equal to x subplot(2,2,1); stem(n,x); title('Rectangular pulse') xlabel('n'); ylabel('x(n)'); axis([-10,10,0,1.2]) subplot(2,2,2); stem(n,even); title('Even Part') xlabel('n'); ylabel('Even of x(n)'); axis([-10,10,0,1.2]) subplot(2,2,3); stem(n,odd); title('Odd Part') xlabel('n'); ylabel('Odd of x(n)'); axis([-10,10,-0.6,0.6]) subplot(2,2,4); stem(n,x); title('Reconstructed Rectangular pulse') xlabel('n'); ylabel('y(n)'); axis([-10,10,0,1.2]) Lab Exercise 2 This code snippet takes convolution of two signals. x = [3, 11, 7, 0, -1, 4, 2] h = [2, 3, 0, -5, 2, 1] y= conv(h,x) subplot(3,1,1); stem(x); title('X signal') xlabel('n'); ylabel('x(n)'); %axis([-10,10,0,1.2]) subplot(3,1,2); stem(h); title('H signal') xlabel('n'); ylabel('h(n)'); %axis([-10,10,0,1.2]) subplot(3,1,3); stem(y); title('Convolution of X and H') xlabel('n'); ylabel('y(n)'); axis([-1,25,-0.6,0.6]) Lab Exercise 3 This part introduces you to the procedure of taking cross correlation between two signals using convolution. Cross correlation tells the level of similarity between two signals. This can be used to detect presence of a particular shape in a given sequence.

x = [3, 11, 7, 0, -1, 4, 2]; nx=[-3:3]; % given signal x(n) & index nx y = [0 0 x]; ny=[-3:5]; % obtain x(n-2) w = randn(1,length(y)); nw=ny; % generate random noise z = y + w; nz=ny; % generate noisy x(n-2) x = fliplr(x); %Prepare x for cross correlation using convolution Xrr = conv(z,x); stem([-6:8],Xrr); title('Cross Correlation') xlabel('index'); ylabel('Correlation'); Lab Exercise 4 This part will use all the above exercises to convolute an exponential signal with a rectangular window. Now we will learn to generate different discrete signals in matlab and convolute them. x = rectwin(51); % to generate a rectangular x[n] of 51 impulses n=0:50; % index for x[n] from 0 to 50 a = 0.95; % exponential factor h = a.^n; % exponential impulse response h[n] subplot(3,1,1),stem(n,x), axis([-10 60 0 1.5]); xlabel('n'); ylabel('x[n]'); subplot(3,1,2),stem(n,h), axis([-10 60 0 1.5]); xlabel('n'); ylabel('h[n]'); y = conv(x, h); N = length(y); m = 0:N-1; subplot(3,1,3), stem(m, y), axis([-10 60 0 20]); xlabel('n') ylabel('y[n]');

Experiment 1: Performing Convolution

Task: Write a function for convolution without using matlab function conv. Generate following discrete signals (x[n]) in MATLAB and convolute them with a rectangular window of any small size. a) A sinusoid with w = 0.02pi. b) A ramp signal. c) A rectangular pulse. Notice the behavior of your system for different values of n and compare.

Experiment 2: Designing a Differentiator System

Task: A differentiator system outputs the difference between current input sample and previous input sample. LCCDE (Linear Constant Coefficient Difference Equation) for this system is y[n] = x[n] - x[n-1] Implement this system as a MATLAB function and give following signals as input to this function. a) A sinusoid with w = 0.02pi. b) A ramp signal. c) A rectangular pulse. Notice the behavior of your system.

Experiment 3: Designing Accumulator System

Task An accumulator system accumulates all the input samples in a register and outputs the sum. LCCDE for this system is y[n] = x[n] + y[n-1] Implement this system as a MATLAB function and give following signals as input to this function. d) A sinusoid with w = 0.02pi. e) A ramp signal. f) A rectangular pulse How much is this system different from the one you implemented in Problem 2 (Differentiator).

IV. LAB 4
Experiment: Noise Removal from a Signal Objective:
In this lab, you will get familiar with designing digital filters using the LabVIEW DSP Module and the Speedy-33 DSP board This lab attempts to explain the frequency domain representation of signals. The frequency domain is simply another way of representing a signal.

Introduction Sinusoidal signals and frequency

Asinusoid is any function of time having the following form:

The following two sinusoids have frequencies of 300 Hz and 500 Hz, respectively:

They can be added together to produce a more complex waveform:

The frequency-domain display shows how much of the signal's energy is present as a function of frequency. What is a filter? A filter is a device that accepts an input signal, and passes or amplifies selected frequencies while it blocks or attenuates unwanted ones. Filters can be analog or digital. The Frequency Response of a filter is the measure of the filter's response (filter output) to a sinusoidal signal of varying frequency and unit amplitude at its input. There are many filter types, but the most common ones are lowpass, highpass, bandpass, and bandstop. They are shown below along with their frequency response.

Part A. Generating Signals Using Sinusoids Procedure i. Choose EMB Sine Waveform from Functions->Embedded Signal Generation ii. Connect 2 control knobs to the amplitude and frequency nodes of the EMB Sine Waveform block. Add a Graph to show the sine wave. iii. Observe the sine wave on the graph while varying the amplitude and frequency knobs. iv. Using the above information, generate a square wave using EMB Sine Waveform.

Part B. Plot Signals in Frequency Domain

Procedure i. In order to display the frequency content of an audio signal, construct first the block diagram. Set the sampling frequency in the input and output to 8000 Hz ii. Add two graphs corresponding to the audio input for the two channels. iii. Change the xlabel name of the two graphs in the Front Panel window to frequency instead of time. Right-Click on both graphs and choose Properties. Under the Scales tab, change the range of the frequency axis to 0-255. iv. To display the frequency content of the audio signal use the Fast Fourier Transform (FFT), go to the block diagram window and connect the Spectral representation of the input signal to the 2 graphs. Look in the Signal Processing sub-palette for the FFT block.

v. Run the VI and notice the frequency range of your voice. Try to whistle and notice the corresponding frequency spectrum. Comment. Change the pitch (frequency) of whistle and examine how it affects the frequency spectrum. Try experimenting with different notes.

Part C. Frequency Response

In this part, you will design a lowpass digital filter and then you will draw its frequency response, in other words you will draw the output amplitude at different frequencies of the input signal

Connection Diagram:

Procedure:
1. Add an EMB Sine Waveform. 2. Insert a filter located on the Functions Signal Processing Filters Filter palette. 3. Configure the filter to have a lowpass filter type with a cutoff frequency of 500 Hz. 4. In the Front Panel, place two knob controls and a Waveform Graph. Name the two knobs Amplitude and Frequency. Make sure to let the Frequency knob vary from 0 to 10000. 5. Go back to the Block Diagram and make all the necessary wiring as shown above 6. Save your VI as lab2-group#-.vi and run it. 7. Set the amplitude of the input sine wave to 6. Vary the frequency of the input sine wave from 10 Hz to 1000 Hz. Observe the output amplitude at each frequency and plot the resulting frequency response. 8. In analog filter design, the cutoff frequency is usually defined to be where the amplitude is reduced to 0.707. Repeat the previous steps for a highpass filter with the same cutoff frequency. 9. Find the frequency at which the output amplitude is 0.707*input amplitude. Does it correspond to 500 Hz? Explain

Part D. Noise Removal

1. In this part, you will have to design your own system to remove noise from a corrupted signal. 2. In order to do so, find the corrupted audio file on your desktop and play it. 3. The file was corrupted by adding a sinusoid tone to it. 4. Connect the output sound card of the PC to the input of the DSP board using the double jack stereo cable. The corrupted audio file is now the input signal for the Speedy-33 board. 5. Find the frequency of the noise by looking at the FFT of the input signal. Hint: In the frequency spectrum, find the peaks with constant amplitudes. 6. Design a bandstop filter to eliminate the noise components. Hint: A lowpass and a high pass filter can be combined to produce a bandstop filter

V.

LAB 5
Applications using LEDs and Switches using the Speedy-33
In this lab you will be implementing an LED VU (VU stands for Volume Units) meter that shows the relative strength of an acoustic signal. Another application will be to simulate a roulette game.

Figure: LEDs and Switches on Speedy-33

Experiment- 1: Controlling LEDs using Switches

Task
As a starting application, you will implement a VI to check the status of the DIP switches and accordingly manipulate the corresponding LEDs; i.e., if DIP switch 1 is ON, turn LED1 ON, and so on.

Procedure
1. Choose from the Functions palette the Digital Input node found in Elemental I/O. 2. By placing the Digital Input node in the window, you can read the status of a selected DIP switch. 3. This can be done by first configuring the Digital Input node to the corresponding DIP switch. By double-clicking on the Digital Input node, the Configure Elemental I/O panel opens up, which allows you to configure the digital input. 4. In the General tab, select On-board DIP switch as a resource.

5. In the Configuration tab, select the DIP switch that you want to read its status. Choose Switch#1 as shown in Fig below. 6. Once the Configuration step is completed, connect the output of the Digital Input node to an indicator, by right-clicking on the output of the Digital Input node and choosing Create->Indicator. 7. Place a While Loop so that the VI keeps running when you press the Run button 8. Try pressing different DIP switches, while watching the indicator in the Front Panel. What happens? Comment. 9. You can configure the LEDs in a similar way: 10. This is done by choosing from the Functions palette the Digital Output node found in Elemental I/O.

11. By placing the Digital Output node in the window, you can control the status of any LED. 12. This can be done by first configuring the Digital Output node to the corresponding LED. By double-clicking on the Digital Output node, the Configure Elemental I/O panel opens up, which allows you to configure the digital output. 13. In the General tab, select On-board LED as a resource.

14. In the Configuration tab, select the LED that you want to read its status. Choose LED#1. 15. Once the Configuration step is completed, connect the input of the Digital Output node to a control, by right-clicking on the input of the Digital Output node and choosing Create->Control. 16. The Front Panel and Block Diagram windows should show a Boolean control. 17. Try pressing the LED control in the Front Panel, while watching the LEDs on the board. What happens? Comment 18. Remove the control from the Digital Output input and connect it to the output of the Digital Input as shown in figure above Try pressing the LED control in the Front Panel, while watching the LEDs on the board. What happens? Comment! Modify the VI so that you check the status of the 8 DIP switches and accordingly manipulate the 8 corresponding LEDs. Run the VI and press different switches and notice the corresponding LED that turns on. Comment.

Experiment 2: Audio LED VU Meter

For audio systems or recorders, an audio VU (VU stands for Volume Units) meter is a device that indicates the relative levels of the audio being recorded or played. In order to implement an Audio LED VU meter using the Speedy-33 board, you should acquire an audio signal and compare its energy level with diferent levels, if the energy is higher than a certain level, a corresponding LED will be turned on.

Procedure
1. The first step to do is to acquire the audio signal by placing an Analog Input node in the Block Diagram window, with a sampling rate of 8 kHz. 2. In order to get the energy level of the signal block for the RMS block, which you can find the Signal Processing > Time Domain sub-palette 3. Connect the output of the analog input block to the input of the RMS block. Connect the output of the RMS block to an indicator, so that you can see the energy level of the incoming signal. The RMS block computes the Root Mean Square (rms) of the input sequence x using the following equation:

4. Next, you should compare the value of the energy level with different constants. If the energy level is higher than a constant, then the corresponding LED should be turned on. 5. In order to do so, find the Greater block found in the Comparison subpalette. 6. Place 8 Greater blocks, each block corresponding to a LED. 7. Connect the output of the RMS block to the input x of the Greater blocks. 8. Connect a constant to the second input y each of the Greater blocks, by right- clicking on the input y and choosing Create->Constant. 9. Place 8 constants for each Greater block. Choose the constants to be 0.3 0.4, 0.5, 0.6, 0.7, 0.8,0.9 and 1 as shown in Fig. 5-20. These constants were found after testing to provide a good level of sensitivity to the audio level meter LEDs. 10. The last step is to compare the signal energy level of the incoming audio signal with a certain level, and turn on a LED depending on the comparison result. 11. In order to do so, connect 8 Digital Output nodes, corresponding to 8 different LEDS, to the output of the Greater blocks. Place a While Loop.

Connection Diagram

Run the VI and comment on the status of the LEDs when you speak. Try to vary the sensitivity of the VU meter by placing a Multiplier function and connecting its 2 inputs to the RMS node output and to a control. The output of the multiplier will be connected to the input of the Greater functions.

VI. LAB 6 Sampling of Analog Signals

Objective:
In this lab you will learn the sampling process and observe aliasing and oversampling happening using Matlab. Exercise: Plot a cosine wave of frequency 0.4 using a fine increment in time scale. Plot the Fourier transform of this signal. t = 0:0.1:100; frq = 0.4; cosx = cos(2*pi*freq*t); plot(cosx) Now plot an impulse train with time period T, to sample this signal. Plot the Fourier transform of this signal. Ts=10; Imp=zeros(1,1001); for i=1:1001 if(mod(i,Ts)==0) Imp(i)=1; else Imp(i)=0; end end plot(Imp) Generate an impulse train of input signal. Plot the Fourier transform of this signal. What output do you see? Your job is to realize the plot. ss=cosx.*Imp; This is our sampled input signal ss.

LAB TASKS:
1- Acquire sampled versions of the following signals. a) Take cosines with different frequencies 0.1, 0.4 and 1.2. b) Take sines with different frequencies 0.1, 0.45 and 1.2. 2- Undertake different sampling rates to show aliasing and over sampling phenomena.

VII. LAB-7
Applications of Correlation Functions
Background and Objective:
The Dolby Surround-system is an audio system, which contains several speakers. They can be placed in different way at the room. The different part of a music piece can be sent to different speakers. Since the sound speed is only about 340 m/s, the sound from different speakers is listened to by listener with certain relative time delay, because of different distances between listener and speakers. If the distances are known, it is possible to compensate the time delays by the way of contrary delays. E.g. if the sound from the speaker achieves the listener 1 ms later then the sound from others non compensated speakers, then to compensate it, the signals sent to the others speakers have to be delayed 1 ms. The task is that you apply sound to measure a distance between speakers and listener in a Dolby Surround-system (audio system using several speakers). The information about the distances simplifies compensation system, which gives opportunity to listen to the sound from all speakers in proper time instant and thereby to make better sound quality. In the system that is applied to measure distance between the speakers and the listener, we assumed that the listener uses the microphone and that the sound is sent back to the music system. The distances can be measured using test signal, which is sent only to one speaker at the time. The delay between the test signal, which has been sent, and the signal, which was registered by microphone, can be estimated by calculating a cross-correlation between the signals. The distance can be calculated just multiplying the time delay by sound speed. In real, you do not need to calculate the distances since the time delay is the interest of the application.

Experiment: Distance measurement Apparatus: Speaker, Microphone, Recording System, MATLAB

Several experiments have been performed to find, which measurement set the best is to measure distance between speakers and microphone by measuring sound delay. The experimental set-up is shown in the Figure. The speaker is placed at the left hand side. The reflection wall is shown on the right hand side. The microphone is placed between the speaker and the wall. The distance between the speakers and the microphone, lhm, has varied. The distance between the speaker and the wall, lhv, was constant. The data for experiment can be loaded. The test signal, u, which has been sent via the speaker, contains three periods of high frequency sinusoidal signal and zero elsewhere. At the same time when the test signal was sent, the signal from microphone was registered. The signal is named y. The microphone was placed at fixed distance from the speaker during the experiment

Tasks
Plot the data of the experiment. What was lhm in experiment? What was lhv? Estimate accuracy of the measures. Estimate crosscorrelation function between u and y for the experiment. It can be done with command: xcorr (u,y,'unbiased') Estimate time delay for the experiment? Prepare a report, adding your work including (matlab code) and your own description.

VIII. Lab 8 Implementation of Digital Audio Effects: Echo, Reverb

Objective
In this lab, you will acquire a working knowledge of the basic tools used in digital audio processing. In particular, you will implement echo effects using buffers to generate delayed playback, and reverberation, which is a simulation of an echoing room.

Echo
Simply put, an echo takes an audio signal, and plays it back after a delay time. The delay time can range from several milliseconds to several seconds.

Figure: Diagram of the single echo effect

Just having a single echo effect is rather limiting, so most delays also have a feedback control (sometimes called regeneration) which takes the output of the delay, and sends it back to the input, as shown in Fig. Now, you have the ability to repeat the sound over and over, and it becomes quieter each time it plays back (assuming that the feedback gain is less than one. Most delay devices restrict it to be less than one for stability). With the feedback, the sound is theoretically repeated forever, but after some point, it will become so low that it will be below the ambient noise in the system and inaudible.

Figure: Diagram of echo with feedback

Reverberation
Reverberation (reverb for short) is probably one of the most heavily used effects in music. When you mention reverb to a musician, many will immediately think of a stomp box, signal processor, or the reverb knob on their amplifier. But many people don't realize how important reverberation is, and that we actually hear reverb every day, without any special processors. The effects of combining an original signal with a very short (< 20ms) time-delayed version of itself results in reverbation. In fact, when a signal is delayed for a very short time and then added to the original, the ear perceives a fused sound rather than two separate sounds. If a number of very short delays that are fed back on themselves are mixed together, a crude form of reverberation can be achieved.

More specifically, reverberation is the natural decay of many delayed versions of an original signal after the original has stopped sounding. It occurs in enclosed spaces as the reflections from the floors, walls and ceiling continue to propagate after the sound source has stopped producing sound.

Shift Registers

Figure: Shift Registers within loops in LabVIEW

While loop and for loop structures can have terminals called shift registers that you use for passing data from the current iteration to the next iteration. Shift registers are local variables that feed forward or transfer values from the completion of one iteration to the beginning of the next. Ashift register has a pair of terminals directly opposite each other on the vertical sides of the loop border. The right terminal, the rectangle with the up arrow, stores the data at the completion of the iteration. LabVIEW shifts that data at the end of the iteration, and it appears in the left terminal, the rectangle with the down arrow, in time for the next iteration. You can use shift registers for any type of data, but the data you wire to each register terminals must be of the same type.

Echo Implementation using Arrays

An array is one of the simplest data structures. Arrays hold a series of data elements. Individual elements are accessed by their position in the array. The position is given by an index. The index usually uses a consecutive range of integers. Some arrays are multidimensional, meaning they are indexed by a fixed number of integers. Generally, one -and two-dimensional arrays are the most common. The first index in LabVIEW is 0.

Figure: Array structure

Although a buffer for delay would consist of several thousand elements, let us consider a simple buffer containing just 8 elements. The elements within the buffer are numbered from 0 to 7. The first element is buffer[0], the second element buffer[ 1] and the final element is buffer[7]. 0s are assigned as initial values to each element. One way to use the buffer as a delay is to save the newest measurement in one end of the buffer and the oldest at the other. For example, we may wish to use buffer[0] to hold the newest input and buffer[7] to hold the oldest. This arrangement is known as a straight bufer. Every time a measurement is made by the input, a new value is put into buffer[0]. We discard the oldest reading at buffer[7] then shuffle all the values along one place. For example, buffer[6] would be moved into buffer[7]. Diagrammatically, the process is shown in Fig. 7-5, assuming that the buffer already contains the values 1,2,3,4,5,6,7 and 8. Unfortunately, this arrangement is not suitable for implementing echo and reverberation. If the buffer is large, say several thousand values, most of the time is spent by the program moving data from one location to another. This does not leave enough time to do the audio processing! A more suitable arrangement is to use what is known as a circular bufer, which will be covered in this lab. Acircular buffer works in a different way. Values within the buffer are not shuffled along. Instead, the oldest element is over-written by the newest. This means that the location of the oldest element in the buffer moves every time a new sample is saved. We keep track of the position of the oldest element with a pointer as shown in Fig. 7-6, which assumes that the buffer already contains the values 1,2,3,4,5,6,7 and 8. In order to insert a new value into the buffer, the oldest value is simply overwritten and the pointer moved along one place. Unlike a straight buffer, all the other values still remain in exactly the same place - they are not shuffled along. When the end of the buffer is reached the oldest value is overwritten with the newest. In this case the pointer is set back to the beginning of the buffer again, as if the buffer were circular.

Application # 1: Echo
As a starting application, you will implement an echo. The echo is one of the simplest effects out there, but it is very valuable when used properly. A little echo can bring life to dull mixes, widen your instrument's sound, and even allow you to solo over yourself.

Connection Diagram

Procedure:
1. The first thing to do is to replicate in Lab VIEW the diagram shown in Fig. 2. The Input and Output blocks are selected from the Functions palette by choosing the Analog Input and Analog Output nodes respectively found in Elemental I/O. 3. Configure these blocks to have 1 channel single sample, 8000 Hz sampling rate and a gain of 1. 4. Next, the Sum block should be added by selecting, from the Functions palette, the Sum node found in the Numeric sub-palette. 5. Connect the output of the Analog Input to the input lines of the Sum block, and the output line of the Sum block to the input of the Analog Output. 6. Next, the Delay block should be implemented which delays the input by a specific amount of time and adds it to the second input of the Sum block. 7. In order to do so, insert an array which will be used to store past inputs. This is done by choosing from the Array subpalette the Index Array. 8. The array should initially contain 1000 elements of 0s. Therefore, add a Simulate Signal VI and double-click on it. Configure the VI to have DC signal type, Offset equal to 0, 8000 samples per second and 1000 samples. 9. Connect the output of the Simulate Signal VI to the input of the array node of the Index Array. 10. The next step is to replace the oldest value stored in the array by the newest data acquired from the Analog Input and shifting the array data by one place. 11. Therefore, add Replace Array Subset from the Array Subpalette. This block will be used to replace the oldest sample stored in the array by the newest sample acquired from the Analog Input. Therefore connect the output of the Analog Input to the new element node of Replace Array Subset. 12. However, the user should be able to control past inputs and attenuate them. Therefore add a Multiply block and connect a control slider and the output of the Analog Input to the two inputs of the Multiply block.

13. The next step is to choose the oldest value from the array and add it to the newest value acquired from the Analog Input so that an echo is generated. Therefore, connect the output of the Index Array to the second input of the Add block. 14. The next step is to increment the pointer to the oldest value as in the circular buffer method explained in the Background section. 15. To do so, add the Quotient & Remainder block from the Numeric subpalette. Connect its x-y*floor(x/y) node to the index of the Index Array block and Replace Array Subset. 16. Add a slider to the y input of the Quotient & Remainder block, which allows you to control the delay. Make sure that the delay range is 0-1000. 17. Knowing that the delay is calculated using the sampling rate and the buffer length, which is controlled by the slider as Delay = Sampling Rate/Buffer Length, what is the maximum delay that can be achieved? 18. In order to increase the pointer to the oldest data, add an Increment block from the Numeric subpalette 19. Since the pointer is incremented in each loop, add a While loop to all the blocks except the Simulate Signal VI, since this one is used once to initialize the buffer to 0s 20. The last step is to shift the values in the buffer circularly. This is done by simply rightclicking on the border of the While Loop and choosing Add Shift Register. 21. Connect the output of the Replace Array Subset to the right terminal, the rectangle with the up arrow. It stores the data at the completion of the iteration. LabVIEW shifts that data at the end of the iteration, and it appears in the left terminal, the rectangle with the down arrow, in time for the next iteration. Connect the latter to the element node of the Index Array, as shown in connection diagram. 22. Repeat the same with output of the Increment block. Initialize the pointer to 0. To initialize a shift register, wire a value from outside the loop to the left terminal. If you do not initialize the register, the loop uses as the initial value the last value inserted in the register when the loop last executed, or the default value for its data type if the loop has never before executed. You should normally use initialized shift registers to ensure consistent behavior. Run the VI and try playing with the delay. Notice the effect on the Echo. Show your Results. Comment on Results.

IX.

Lab 9
FIR and IIR Filter Design Techniques

Objective:

Objective of this lab to design FIR, IIR Filters and to compare the two filter types

Exercise 1 FIR Filter Design Procedure: Order=15; Wn=0.4; % Cut-off frequency b=fir1(Order,Wn) % See help [h,q] = freqz(b,1,512); subplot(2,1,1) plot(q/pi,abs(h)); title('Magnitude Response'); subplot(2,1,2) plot(q/pi,phase(h')); Order=50; Wn=0.4; % Cut-off frequency b=fir1(Order,Wn) % See help [h,q] = freqz(b,1,512); figure subplot(2,1,1) plot(q/pi,abs(h)); title('Magnitude Response'); subplot(2,1,2) plot(q/pi,phase(h')); Responses should look like this
Magnitude Response, Phase Response of first filter (Order= 15) Magnitude Response and Phase Response of first filter (Oder = 50)

Exercise 2 IIR Filter Design Procedure: Order=15; Wn=0.4; [b,a]=butter(Order,Wn); [h,q] = freqz(b,a,512); figure subplot(2,1,1) plot(q/pi,abs(h)); title('Magnitude Response'); subplot(2,1,2) plot(q/pi,phase(h')); Wn=0.4; [b,a]=butter(Order,Wn); [h,q] = freqz(b,a,512); figure subplot(2,1,1) plot(q/pi,abs(h)); title('Magnitude Response'); subplot(2,1,2) plot(q/pi,phase(h')); LAB TASKS: Write code for the following tasks using Butterworth, Cheby1, Cheby2 and Elliptic filter. A- Design corresponding high pass filters for the low pass filters constructed above. B- Design a Band Pass filter with wl = 0.2 pi and wh = 0.5 pi. Lab Assignment: You have been given a sound file i.e. guitar.wav. Add noise to this signal at different frequencies to generate a noisy signal. Design a filter for each case so that noise is eliminated. To check play the sound of final signal (It should be same as original guitar.wav). 1. Add a noise of 50 Hz to the audio signal. 2. Add a noise of 500 Hz to the audio signal. 3. Add a noise of 50 Hz and 500 Hz to the audio signal.

X.

Lab 10 Design and Test of Music Equalizer

Music Equalizer
Frequency is the number of vibrations or oscillations second, which is measured in cycles per seconds or Hertz. It is said that the human ear can perceive a range of sound frequencies from 20 Hertz (twenty cycles per second) to 20 kiloHertz (twenty thousand cycles per second), but of course, that does not apply to all of us. Audio equalizers allow you to adjust levels at set of points in a range of frequencies.

Connection Diagram

i. After starting LabVIEW Embedded DSP Module and changing the target to the SPEEDY-33 board, create a New Blank VI. 1. Open the Block Diagram and place the Elemental I/OAnalog Input node in it. ii. Double click the Analog Input node and change the Resource under the General Tab to be 1 channel multiple samples and check that the Sample Rate is set to8000 Hz and Frame Size is 512. 1 channel means that the output is mono and multiple samples means that more than one sample is taken each sampling time and is stored in a buffer size of 512 in our case. Repeat the same for the Analog Output node. iii. Place a Spectral Measurements Express VI and double-click on it to select Power Spectrum from the type of Measurement. iv. Rearrange the objects in the Block Diagram so that you have plenty of space between them. Next, the incoming signal should be partitioned into different bands, so that one can control different frequency regions. The more regions available for adjustment within a frequency range, the more precise the audible adjustments become. In this experiment, 5 regions will be used: 0-50 Hz, 50-100 Hz, 100-250 Hz, 250-750 Hz, 7001500 Hz, >1500 Hz.

v. Double click the Analog Input node and change the Resource under the General Tab to vi. To do that, in the block diagram, place the Signal ProcessingFiltersFilter VI. To choose the frequency range 0-50, double click in the VI and configure a lowpass FIR filter with the following settings: Type: Lowpass Sampling Frequency: 8000 Hz. Cutoff Frequency: 50 Hz IIR filter: Elliptic, Order 7 vii. The Elliptic type and Order 7 were chosen since they would result in a sharp filter but any other type and order could still be chosen. viii. Connect the output of the Analog Input to the input of the Filter. ix. Repeat the same for the other bands, using a lowpass and a highpass filter in cascade with the lower frequency of the band being the cutoff of a highpass filter and the higher frequency of the region being the cutoff of a lowpass filter. For instance, for the region 50100 Hz, use a highpass filter of cutoff 50 Hz and a lowpass of cutoff 100 Hz in cascade. It is easier to use a bandpass filter; however, LabVIEW DSP Module does not support with Filter VIs bandpass filters. For the last region, you should use a highpass filter with cutoff 1500 Hz. x. The set of 10 filters splits the analog signal into the 6 frequency bands shown on the front panel. The top filter is a lowpass filter that passes frequencies from 0 Hz to 50 Hz. The four pairs of lowpass-highpass filters create bandpass filters of different ranges. The bottom filter is a highpass filter that passes frequencies above 1.5 kHz. xi. The next step is to control the output of each filter by either boosting the input signal to the filter or attenuating it. In order to do that, go the Front Panel and place 6 slide controls in the Front Panel. Place 6 slide indicators under each slide control. You should also add an additional knob control for the volume. Set the range of all slide controls to be between 0 and 2. xii. Put 6 Multiply functions and wire the output of the filters to them along with the output of the Volume control. Put another 6 Multiply functions and wire the output of the previous multiply functions with the Frequency Slide Controls as shown below. The Multiply functions are used to increase or decrease the volume and amplify or attenuate the sound in a specific frequency range. xiii. Add the 6 signals together to get the signal at all frequencies, by placing 5 Add functions and wire the output of the last Add to the Analog Output node and the signal input of the Spectral Measurements Express VI. xiv. Next, the RMS value of the signal at a frequency within each frequency band should be displayed using the slide indicators on the front panel. xv. In order to do that, add an RMS block and place 5 index arrays from Arrays>>Index Array in the Block Diagram. Connect the output of the Spectral Measurements to the array of each of the 5 Index Arrays and the output of each array to the corresponding slide indicator, as shown in Fig. xvi. In order for each Index Array to take the RMS value of the signal at a frequency within each frequency band, the index of each Index Array should be 2,4,8,16,64 and 128. After placing a While Loop, the final diagram should look as shown in fig above.

Start your CD. Save and Run the VI. You should hear the music basically unfiltered, since all frequencies in the range 0-4000 Hz are passed without attenuation. Start playing with the individual Gains now to boost/attenuate the various frequency ranges by varying the gains.

XI.

LAB 11 Generation of a DTMF signal A Telephone application

In this experiment, you will generate a dual-tone multi-frequency (DTMF) signal, which is a signal consisting of the sum of two pure sinusoids at valid frequencies. Touch tone telephones use DTMF signals.

Connection Diagram

How does a telephone work?

Telephone touch pads generate dual tone multi frequency (DTMF) signals to dial a telephone number. When any key is pressed, the tones of the corresponding column and row are generated, hence dual tone. As an example, pressing the 5 button generates the tones 770Hz and 1336Hz summed together. The frequencies in were chosen to avoid harmonics. No frequency is a 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 dial signal in the presence of line distortions.

DTMF generator Procedure

1. In this part, you will simulate touch tone telephones, which use DTMF signals. Dual-tone multi-frequency (DTMF) signals are signals that consist of the sum of two pure sinusoids at valid frequencies. 2. In order to generate a DTMF signal, open LabVIEW DSP Module and create a blank VI. 3. In the Block Diagram window, place two Simulate Signal Express VI located on the Functions Embedded Signal Generation palette. 4. Add the output of the two Simulate Signal VIs and connect the result to the Analog Output (also known as Codec Out). 5. Next, the frequency of each signal needs to be specified. Add two Index 6. Array VIs from the Array subpalette by adding two Index Array VIs. An array, also known as a vector or list, is one of the simplest data structures. Arrays hold equally-sized data elements, generally of the same data type. Individual elements are accessed by index using a consecutive range of integers. They can be one-dimensional or n-dimensional. In LabVIEW, the first index in an array is 0. 7. Then, right-click on the array node of the two Index Array VIs and specify as constant. 8. Next, the possible frequencies should be added into the arrays. The upper array is used to generate one of the low frequency values 697, 770, 852 and 941 Hz, while the lower array will generate one of the upper frequency values 1209, 1336, 1477 Hz. For example, when pressing the button 1, the dual tone generated has the two frequencies 697 and 1209 Hz . Therefore, the element corresponding to index 1 of the upper array will hold 697 while the one in the lower array will hold 1209. Fill in the rest of the array elements by changing the index n-dimensional array according to following Table. Table

8. 9. 10. 11. 12.

Connect the Index Array output to the Frequency node of the Simulate Signal VI. Next, two frequencies should be selected depending on the telephone keypad touched. In order to do that, in the Front Panel window go to Boolean>>OK Button and place the buttons in a way to get the pattern. Go back to the Block Diagram window, and arrange the blocks for clarity: The final step to do is to generate a dual tone by pushing one of the 12 buttons. Therefore, each button push should correspond to selecting a specific element from the two arrays to specify the two frequencies of the dual tone signal. In order to do that, add a Select block from the Comparison subpalette and connect the output of the OK Button 5 to the s input of the Select block.

If button 5 is pushed, then the elements corresponding to index 5 in the upper and lower arrays should be selected. Therefore connect a constant of value equal to OK Button # to the t input of Select block, which corresponds to the event of pushing button 5 and a value of 0 to the f input. Add all the outputs of the Select buttons to each other and connect the result to the index of the arrays: In order to generate the tone for a certain period of time, one should loop the sine generation for different iterations, depending on the desired tone length. Analog Output usually determines the execution speed for the code inside the For Loop, using the following formula: Where #Samples and Sampling Rate are determined in the Simulate Signal block Properties. For example, if #Sample is 128, and Sampling Rate is 8000Hz then the For Loop takes 16 msec for 1 iteration. Add a For Loop around the Simulate Signal blocks along with the Analog Output block. When dialing a telephone, the minimum tone length should be 40 ms. We will choose to generate 150 ms. Knowing that each For Loop is 16 ms as mentioned previously, the number of iterations required will be the 1 50(ms)/1 6(ms). Use a Divide block to compute the quotient and use the result as the number of iterations. Try pressing different buttons and notice the tone generated with each one. Compare it with the tone generated from a touch tone phone. Comment. Try dialing a number using the VI by pressing the corresponding buttons and holding the speaker next to the regular touch phone. Comment.

XII.

LAB 12

Z-Transform Operations

1- Polynomials and Partial Fraction Expansion Partial fraction expansion of polynomials and roots of the polynomials are important while determining the z-transform and its inverse. MATLAB provides with the following functions. a) residue b) poly See help for these functions in MATLAB and do the following example. b = 1; a = poly ([0.8, 0.8, -0.8]) [r,p,k] = residuez(b,a) [b,a] = residuez(r,p,k) The first call to residuez with 2 arguments performs partial fraction expansion of the polynomial given by x and y. The second call with 3 arguments reverses the operation. The following example shows how to calculate magnitude and phases of the complex roots. b = [1,0.4*sqrt(2)] a=[1,-0.8*sqrt(2),0.64] [R,p,C] = residuez(b,a) Mp=abs(p') % pole magnitudes Ap=angle(p')/pi % pole angles in pi units 2- Pole Zero Plot MATLAB provides a suitable way of plotting poles and zeros of a transfer function (polynomial) in the z-plane. (Look for help for zplane in MATLAB). The zplane function can plot poles and zeros for transfer function polynomials as well as explicit roots. Try both versions. Check carefully the difference between the two modes and the set of arguments and their arrangement in row or column wise manner.

b = [1,0]; a = [1, -0.9]; [H1,H2,H3] = zplane(b,a); set(H1,'markersize',10,'color',[0,1,0]); set(H2,'markersize',10,'color',[1,0,0]); title('Pole-Zero Plot'); text(0.85,-0.1,'0.9');text(0.01,-0.1,'0'); figure, [H1,H2,H3] = zplane(b',a'); set(H1,'markersize',10,'color',[0,1,0]); set(H2,'markersize',10,'color',[1,0,0]); title('Pole-Zero Plot'); text(1,-0.1,'1');text(0.01,-0.1,'0'); text(-0.95,-0.1, '-0.9')
Pole-Zero Plot 1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -1 -0.5 0 Real Part 0.5 1 0 0.9

z-plane plot for polynomial z/(z-0.9)

Pole-Zero Plot 1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 -1 -0.5 0 Real Part 0.5 1 -0.9 0 1

z-plane plot for polynomial z(z-1)/(z+0.9)(z-1)

3- Z-Transform and MATLAB MATLAB can take z-transform of a polynomial for you. For this you need to define the independent variable n as a symbol in MATLAB. See help for syms for defining symbols. syms n syms a ztrans(n^2) ztrans(n*a^n) ztrans(a^n) ztrans(n*((0.5)^n)*cos(pi*n/3))

Lab Tasks:
1- Apply partial fraction expansion on the following System using MATLAB. Draw pole zero plot.
H ( z) = 1 + 0 .5 z 1 (1 + 0.5 z 1 )(1 + 0.25 z 1 )

2. Draw pole zero diagram for the following pair of poles. Poles: 0.14, 0.5 Zeros: 0, 0.1 3. Draw pole zero plot using tf2zp

XIII. LAB 13 Designing a Modem

This lab introduces Frequency Shift Keying (FSK), which is used in modems and which is the process of frequency modulation in which the modulating signals shifts the output frequency between two predetermined frequency values. The DSP target produces an FSK signal by generating, through the DAC output channel on the DSP target, a 400-Hz sinusoid to represent a zero or a 700-Hz sinusoid to represent a one.

Modem: Modulator
As already mentioned, the data sent to a modem is 0 or 1 since it is digital information. Therefore, the data will be simulated using a random number generator of 0 and 1. Connection Diagram:

Procedure:
1.In order to do that, open LabVIEW DSP Module and choose the execution target to be Lab VIEW for Windows. 2.In the Block Diagram window, insert Random Number from the Numeric 3.subpalette. Switch the execution target to Speedy33 by going to Operate->Switch Execution Target and selecting Speedy33. 4.The Random Number block generates a number between 0 and 1. In order to generate either 0 or 1 with equal probability, one should compare the output of the Random Generator with 0.5. If the generated number is greater than 0.5, than the data to be sent is 1, otherwise it is 0. Add the appropriate blocks. 5.The output of the Greater block will be Boolean, where True represents a 1 and 6.False represents a 0. If the data to be sent is 0, then the frequency of the sine wave to be transmitted is 400 Hz. Otherwise, it is 700 Hz. Add the appropriate blocks. 7.Connect the output of the Select block to the frequency of the sine wave simulated using the EMB sine wave block. The sine wave amplitude should also be controlled by using a slider between 0 and 5000. 8.The final step is to send the sine wave generated to the speaker by connecting the output of the EMB sine wave block to the speaker.

9.After adding a While Loop, run the VI and observe the transmitted time domain waveform. 10. Observe the frequency spectrum of the FSK signal by adding the appropriate graph. Comment.

Modem: Demodulator Connection Diagram

Procedure:
1. The demodulator should convert the analog signal to binary digits depending on the frequency of the sine wave signal. If the frequency is 400 Hz, then the bit is 0. If it is 700 Hz, then the bit is 1. 2. The signal is received through the Microphone line and then its content at a 3. specified frequency is analyzed using the Goertzel VI found in Signal 4. Processing>Transforms>Goertzel. The two frequencies of interest are 400 Hz and 700 Hz. 5. If the energy of the signal at 700 Hz is higher than the energy at 400 Hz, then the data received is 1. Otherwise, it is 0. 6. After adding a While loop, run the demodulator VI simultaneously with the 7. modulator running on a different board, while connecting the Speaker line of the modulator to the Microphone line of the demodulator. 8. Observe the transmitted time domain waveform by adding the appropriate graph. Compare with the transmitted bits.

9. Add another graph to observe the frequency spectrum of the demodulated signal by adding the appropriate graph. Comment