Electrical and Computer Engineering (ECE) Projects

The purpose of these projects is to tie together concepts and applications between Linear Algebra and Signals.

Project #1: Fourier Series and Matlab  due Monday November 17

Your task is to prepare as a group a paper covering the following topics:

1. the fundamentals of Fourier Series and the Fourier Transform, and how you see those relating to signals. This should be in your own words, not lifted from some web site or

book, verbatim. Any symbols you use you should be able to explain.

2. mathematics showing why the Fourier Transform is called linear

3. visit, investigate and comment on what can be learned from this interactive site.

4. technically describe what .WAV files are

5. study the basics of Matlab. You may wish to use the following link which was developed specifically to help relate linear algebra, signals and Matlab at WPI.

6. You need something to try Matlab out on. Here is a “mystery tuning fork”. Download it and use Matlab to determine its frequency. Include a copy of your Matlab worksheet with the paper you turn in. Bonus points for determining

the length of time recorded and the sampling frequency. (the easiest way to

download it is to right-click on it)

include all references (people, books, web url’s) that you use and have a cover page listing the names of the group members, the group number, and project number.

Suggestion: divide up the work among your group.

Project #2 The Human Voice

It is assumed at this point that you can perform some of the essentials of Matlab and Signal Processing such as reading .wav files, graphing them and performing the FFT on them. In this project we apply these capabilities to the human voice. This will be done by recording people speaking vowels into .wav files and then analyzing and comparing them.

You will need a microphone which can plug into the sound card on a PC. If you do not have one, there will be one on reserve in the Library to borrow.

Pick 4 people; at least one of each gender. Record each of them pronouncing the vowels a,e,i, o and u. Put each vowel into a

single .wav file.

“How do I record into a .wav file?”. You go into Windows/Accessories/Entertainment/Sound Recorder. You may need to check sound levels to make sure you are really recording anything. Once you get them correct, be careful to quickly start and end the recording for each vowel as .wav files grow rapidly in size. Also you should be able to play them back and hear them!

Suggestion: have a system for your file names as you will generate some 20 files. Something like “Mike_E.wav” perhaps.

Once you have them recorded, begin analyzing them.

for a single person, looking at all 5, what characteristics can you spot that are of interest?

looking at the letter a for all 4 people, are there features that distinguish one person from another?

you will want to employ the FFT in your analysis.

Summarize your analysis; what general conclusions about speech can you draw?

Project #3 Applications - Chose One of the Following

At this point, you have some knowledge of Signals, of Fourier Analysis and some capability with Matlab. Let’s do something with it all.

3A. Voice Print Identification

The fundamental question here is: can one analyze a person’s voice and uniquely identify them? If so, how? If not, why not?

The range of applicability of such a capability is enormous! One could unlock a door with their voice. A handicapped person could access all sorts of service. The government could decide if it was really Osama Bin Laden on an audio tape. But this all requires a high degree of accuracy. For example, if someone were accused of making harassing phone calls and taken to court, would audio identification of them hold up?

Your answer should be based on more than random opinion. You have gathered voice data in Project 2 and you have Matlab for analysis of it. While all conclusive research is impossible under the circumstances, none the less you can look at the data you have and the tools you can use, and either arrive at a scientifically based conclusion or indicate what further evidence you would need to gather to do so. When engineers render opinions, they always provide substantiation for them – this small project is no different.

3B. An Intelligent Hearing Aid

The idea for this project comes from DEKA. It is simple in concept but challenging to implement. A standard hearing aid is basically a miniature receiver, amplifier and speaker, if you think about it. The goal here is to decide if one can filter out “noise” and improve the quality of what the person is hearing.

The challenging part is to recognize what is “noise” is. How do you distinguish signals which are part of, say, the voice of a person speaking to the hearing aid wearer and which are part of background noise from the environment?

You have a number of files of people speaking from Project #2 to study as a starting point.

Related challenge: how do you install a filter in Matlab?

Let’s assume for sake of discussion that the hearing aid envisioned could contain microcircuits that could execute any software you might generate with Matlab (in other words, we will leave that problem for the hardware people to solve)

As a product of your work, you might generate two files – an original of some speech and a modified, after your software has improved it. These should be .wav files and we should be able to play both as well as see the plots of both to experience the fruits of your labor.

3C. Comparison and Analysis of Musical Instruments

Once you can use Matlab to analyze .wav files, you can begin to explore lots of sounds in detail otherwise not possible. For those interested in music, comparison of instruments, or analysis of single instruments can be fun and interesting.

For example, do you have a guitar? We have a number of files recorded from another one (electric). Compare them and see what differences you can find. Samples of the analysis of our guitar can be found at this link.

Once you have finished comparing them, can you speculate as to what engineering decisions or compromises lead to the differences you

found? Could any of the differences be improved by smart electronics, as opposed to expensive manufacturing?

Do you play piano? We have files on the Matlab site that were recorded from a piano on campus. You can analyze them to decide if the piano is in tune or what needs to be done to improve its state of tune. We can probably set you up to actually tune the piano if you feel inspired! This link is to a Java applet that allows

you to play an on-line piano and observe the frequencies of the various keys – handy if you don’t happen to have a piano in your backpack!

Do you play drums? One project would be to take cymbals from two major manufacturers and compare them to see what subtle distinctions there are (I can provide ride cymbals made by Zildjian and Paiste, for example).

Got an idea for something else?? let’s hear about it!