Description
1. Generate two sinusoids at two different frequencies ω0 and ω1 (512 samples). Convolve the two signals. Compute DFT of the convolved signal.
Plot the spectrum. Compute the DFT of each of the signals. Multiply the
spectra, and compare them with that of the convolved signal. Remember
that the DFT order should be >= lengthof thesignal.
2. Record isolated vowels from both yourself and your partner. Compute the
DFT spectrum. Find the first two prominent peaks, say F1, F2. Plot F1
vs. F2.
3. Record unvoiced (UA, A) /k/ /tch/ /t/ /th/ /p/ and their voiced counterparts. Produce all the sounds by suffixing with vowels /a/, /i/ and
/u/. What differences do you see in the formant frequencies of the vowels
in comparison with that of the isolated vowels?
4. Record diphthongs /ai/ and /ao/. Plot the spectrogram. Compare the
spectrograms with that of the individual spectrograms of the original vowels. Experiment with window sizes that include just one pitch period vs.
another that includes at least two pitch periods.
5. Record fricatives /s/ and /h/ by suffixing with vowel /a/. Compare the
spectrograms, and record the same.
6. Generate a set of tones in the range 350-440Hz (fixed amplitude). Make
a composite signal by adding all these tones. This is called the masking
noise. Next generate a fixed frequency tone from 100 Hz to that of 3000
Hz. Change the amplitude of the tone until the tone is just discernible.
Plot the tone amplitude as a function of frequency.
7. Use Lab 1 data with marked vowel boundaries. Estimate F1, F2 for the
vowels. Plot F1 vs F2. Compare with the vowel frequencies obtained in
Question 2.
1
