Homework 3

(points are presented in brackets)

Please submit your homework in the following format:

lastname_firstname.pdf
Please include the names of the students of this group in the document.
Start by downloading and unzipping the file homework_03.zip.

Exercise 1

(4 points)

Connect your headphones to your computer. Then run the file exercise01.m in Matlab. This file
creates a sinus sweep y1 which increases its frequency from 100Hz to 1000Hz in 6s.
Compare y1 and y2 auditory and via their spectrograms (the sweeps are saved as y1.wav and
y2.wav).
- Why do y1 and y2 sound different?
- What can you see in the spectrograms?
- What is the name of this effect and where does it come from?
- What could you do to avoid this effect?
(Take a look at the Matlab-code to understand the differences of both signals)

Exercise 2

(2 points)

Start Matlab and run the file exercise02.m. The hist command creates a histogram which shows
the distribution of data values in the signal.
What does this plot tell you about the signal? Thus, what should you keep in mind when you quantize
a speech signal?

Exercise 3

(2 points)

Note: Do not listen to the wav-files in advance!

As you have learned in the lecture you can identify vocals via their first two formants.
Open the files sound1-5.wav (you can find them in the folder wav) in PRAAT, create the spectrogram
(click on View&Edit) of each wav-file and display the formants (click on Format Show Formants).
Click on Formant --> Formant Settings and check if the following values are set:
Maximum formant (Hz): 4500
Number of formants: 5.0
Window length: 0.025
Dynamic range (dB): 30.0
Dot size (mm): 1.0
Use PRAAT to get the values for F1 and F2. Then use the formant maps on page 51 (German
script)/page 53 (English script) of the script to identify the vowels.
Create a table with four columns: two for the values of F1 and F2 for each sound file and the third
one for the vowel you would guess by using the formant maps. Then listen to the wav files and insert
the actual vowel into column four.

Exercise 4

(3 points)

When humans produce sound they use different parts of their vocal tract, e.g., a sound can be labial,
dental, alveolar, palatal, or velar. Consonants are classified by these places of articulation and their
mode of excitation (voiced, unvoiced).
Create a table and write down two consonants for each of the above mentioned places of
articulation. One of these consonants should be voiced the other unvoiced. Please add one example
word for each sound.
Your solution should look like this:
Labial (lips): voiced -> sound w in wind
unvoiced -> sound f in foot
(Of course you can do this exercise in German)