Description
1. Record your name and the utterance should be ”I am ”. And save it as
a wave file. Perform phone level transcription on the wave file. (Plot in a single figure
window) Note: Computer Based Question [2.5 pts]
(a) 2D plot of STFT based Spectrogram (preferable Matlab). A sample snippet for
output is attached Figure: 1
(b) 3D plot of STFT based Spectrogram (preferable Matlab). A sample snippet for
output is attached Figure: 2
2. Load a speech signal of your interest and select voiced region then Find the pitch based
on the Cepstral analysis? Plot the results (Speech, Cepstrum, Liftered signal;) Note:
Computer Based Question [2.5 pts]
3. Take a recorded speech signal which you have done earlier (Your name from Question 1).
Consider a voiced frame and unvoiced frame of your choice, apply the following. Note:
Computer Based Question [7.5 pts]
(a) no. of points in N-point DFT (256,512,1024)
(b) size of window
(c) shape of window (Rectangular, Hamming, Hanning) on STFT magnitude spectrum.
And comment on each (3(a),3(b) and 3(c))
Figure 1: Output of 1(a) should be in this way
Figure 2: Output of 1(b) should be in this way
Figure 3: Output
4. Take a recorded speech signal which you have done earlier (Your name from Question
1). And extract MFCC for every 20 ms and save in txt file. Plot accordingly as shown
below Figure. 3. And comment on the plot with proper explanation . Note: Computer
Based Question [7.5 pts]
Note
1. Question 1(a), 1(b), 2(a), 3(a), 3(b), 3(c) and 4 are Computer Based Question so you are
expected to code it in MATLAB(preferably) or PYTHON.
