Auralization
Finally, the last task is to auralize our results.
Before being able do that we must not forget that we are still in 1:8 scale. So we have to come back to the original frequency range of interest which would be 800Hz to 16kHz. For that we compensate the spectra with respect to the model. The data are presented in third octave band.
An acoustic guitar song and a speech have been auralized to hear the effects of each device (barrier, wall). Using a filter on those wav files, new mp3 files were created after convolution with our data. We had to filter also the original sound, setting at 0 dB for all frequencies, in order to judge the frequency responses and not the filter itself.
The wall data are given in the ATA book chapter 10 (no scaling has to be considered for this one). One can already guess that the wall should have a better insertion loss since this one is bigger (taller and wider).
While listening to the sound files, one can distinguish easily the original sound and the one filtered by the wall. However the sound filtered by the barrier does not give any audible differences. According to the previous section there should be some difference due to the insertion loss of the barrier. This should happen especially in high frequencies.
However the wall seems to work in the opposite direction. The sound is perceived as high frequency. The low frequencies seem to have been deviated. This is more obvious on the guitar sound file. The barrier, which can be regarded as a thin and short wall, has this effect too but in a less audible way.
All this goes against what has been found in the previous section. Indeed according to the barrier insertion loss there should be less energy in high frequencies than in low frequencies compared to the case without barrier.