Speech
For the maximum signal power frame, 8.9756% of the signal remains.
For the 1/4th max signal power frame,25.8537% of the signal remains.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Project 3: Speech |
|||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||
| Click to hear Speech Sample | |||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||
| 1. The input signal with a 2048 Hann window and 50% overlap (updated after trying 1/16th) and converted to spectral representation via DFT. | |||||||||||||||||||||||||||||||||||||||
| Click to see the Spectrogram | |||||||||||||||||||||||||||||||||||||||
| 2. Compute the 25 critical band Bark spectrum. | |||||||||||||||||||||||||||||||||||||||
| Click to see the Spectrogram | |||||||||||||||||||||||||||||||||||||||
| 3. Select two signal frames, one at the location of the maximum signal power, and the other at 1/4th the maximum power. Plot a) power spectrum, b) Bark spectrum, c) spread Bark spectrum, d) offset Bark spectrum, e) final threshold. | |||||||||||||||||||||||||||||||||||||||
| Click to see Question 3. a) - d) for Max Power | Click to see Question 3. a) - d) for 1/4th MAX Power | ||||||||||||||||||||||||||||||||||||||
| 4. Compute and plot the absolute threshold of hearing. | |||||||||||||||||||||||||||||||||||||||
| Click to see Question 3. e) and Question 4 for Max Power | |||||||||||||||||||||||||||||||||||||||
| Click to see Question 3. e) and Question 4 for 1/4th Max Power | |||||||||||||||||||||||||||||||||||||||
| 5. Determine the signal spectral components that remain audible after masking and falling below the threshold of hearing. For the maximum signal power frame, 8.9756% of the signal remains. For the 1/4th max signal power frame,25.8537% of the signal remains. |
|||||||||||||||||||||||||||||||||||||||
| 6. Reconstruct the time signal with the remaing audible power spectrum. | |||||||||||||||||||||||||||||||||||||||
| I had a difficult time doing this. I tried to follow the tips given to the class to aid in this process, but it wasn't able to successfully reconstruct the signal. Below is a link to the spectrogram I was able to produce (it is obvious the signal is not correct). Along with a sound file that compresses the audio sample with another codec that I didn't create. I am including this to provide an example of what my compressed signal should sound like and would be expected to sound like if I was able to successfully reconstruct the signal. | |||||||||||||||||||||||||||||||||||||||
| Click to see the Spectrogram | |||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||
| Click to hear the compressed speech sample | |||||||||||||||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||||||||||||||
| Click for next page | |||||||||||||||||||||||||||||||||||||||