ELEC 241 Lab

Experiment 5.2

The Spectrum of Acoustic Signals

Part 1: The Spectrum of Speech Signals



Step 1:

 Connect the power supply to your breadboard (if you haven't already done so), plug the dynamic microphone into J1-6, and verify that the mixer circuit you build last week still works.

Step 2:

 Connect CH1 of the scope and A/D input 4 to $v_{out}$ of the mixer circuit.

Step 3:

 Set the spectrum analyzer to dB magnitude scale and Linear frequency scale.

Step 4:

 Produce a sustained vowel (a, e, i, o, u) sound. Note the form of the spectrum. It should consist of a series of peaks, like the signals from the function generator, but unlike the function generator signals where the harmonics fall off monotonically, the amplitudes of the harmonics in a speech signal rise and fall with increasing frequency.

To make it easier to study the spectrum, you can "freeze" the display by pressing the STOP button (continue producing the sound until the freezing is complete). If you like, you can get a printout of the display by selecting "Print Window" from the File menu.

Step 5:

 Estimate the fundamental frequency (pitch) of the waveform. Notice that unlike the function generator signals, the fundamental is not necessarily the strongest component.

It may be difficult to get an accurate estimate if the fundamental is low in frequency (e.g. a male voice). One way to improve accuracy is to find the frequency of one of the higher harmonics and divide by its order.

Step 6:

 While trying to keep the same pitch, produce different sustained vowel sounds. Note how the "envelope" of the spectrum (the line connecting the tips of the peaks) changes while the positions of the lines remains the same.

Step 7:

 Now, produce the same sustained vowel sound (your choice) while varying the pitch. Notice how the overall shape of the spectrum remains the same as the lines move up and down.

Step 8:

 If you brought your musical instrument with you, play a few notes into the microphone. Note the nature of the spectrum.

Step 9:

 A soft whistle should be very close to a pure tone. Whistle a tone of about 1 kHz into the microphone and see if this is the case.

Step 10:

 Plug in the telephone handset. Verify that the carbon microphone input to the mixer still works by speaking into the handset microphone.

Step 11:

 Whistle the same note into the handset microphone and observe the spectrum display. Does it have the same harmonic content as with the dynamic microphone? Since the acoustic signal was the same, the difference must be caused by distortion in the microphone. Which microphone has the least distortion?

Step 12:

 Blow into the microphone and note the shape of the spectrum. This is a broadband, random signal.

Step 13:

 Produce an 'sh' sound. Note the shape of the spectrum. This is called an unvoiced fricative.

Step 14:

 Now try a 'zh' sound (a voiced fricative). How does its spectrum compare with the one in the previous step?

Question 1:

 Based on your observations of different speech sounds, can you determine an approximate bandwidth for speech? I.e. is there a frequency below which "most" of the energy of the speech signals is concentrated?

Part 2: Analyzing an Unknown Signal



Step 1:

 Plug the sound card cable from the Lab PC into P2-( on the second interface module.

Step 2:

 We want to be able to hear the signal, so connect one side of the handset earpiece to the mixer output. Ground the other side of the handset earpiece.

Step 3:

 To play the sound card through our mixer, we will need to add a third input. To do this, add a 100 k$\Omega$ resistor ($R_3$ ) to the mixer circuit so that it looks like this:
\includegraphics[scale=0.550000]{ckt5.2.ps}


Step 4:

 Connect the sound card line out (pin 20 on the interface board socket strip) to $v_{sound}$ .

Step 5:

 Unplug the dynamic microphone from J1-6. Unscrew the cover from the mouthpiece of the telephone handset, and remove the microphone cartridge. Now the only input to the mixer is from the sound card.

Step 6:

 Play the Mystery Signal from Lab 2. You should be able to hear it through the earpiece and see it on the spectrum analyzer display. If necessary, run the Volume Control program and adjust the main Volume Control level control to get a satisfactory signal amplitude.

The first time you play the sound, Netscape pops up a control box that looks like this:

This allows you to replay the signal (the button with the triangle) and adjust the volume (the slider). If you replay the signal by clicking the link in Netscape, you get a new sound box each time you click.

Step 7:

 It may be easier to see what's going on if you set the spectrum analyzer display to linear magnitude and log frequency.

Question 2:

 Based on what you see on the spectrum display, can you explain how the mystery signal works? Hint: focus on a single peak and note how it changes from one tone to the next.

Step 8:

 Put the telephone handset back together.

Step 9:

 Exit the spectrum analyzer program by clicking the "X" box in the upper right hand corner of its window. Answer "No" if the question "Save changes to Spectrum Analyzer 2.vi" appears. Disconnect the ribbon cable from the interface board by using the ejector levers. Do not try to remove it by pulling on the cable.