Spanish Phonemes 

The objective of this experiment, is modeling the sound produced by a running faucet into a bottle (tube). Let us suppose the running faucet generates a type of Gaussian noise in the bottom of the bottle, which is filtered by the acoustic enclosure formed by the walls of the bottle. As the bottle is filled, the acoustic enclosure gets shorter.

Tubes can be modeled as bank of resonant filters, whose center frequencies are given by

Where Fk is the formant number k , c the speed of sound and l the length of the tube. As the length of the tube changes, the bottle is filled. Then, the length can be modeled as a function of time:

If the length of the tube is in centimeters, which is its original length? Of course, 17.5(cm). This number can gives us a clue about the lowest frequency, which is F1(0)=500(Hz).

The following audio examples are the result of this synthesis process. The duration of each audiofile is 5 seconds.

The quality factor is a rate between Fk(t) and the bandwidth of each resonator filter. Several outputs were turned out in order to show the difference using different bandwidths, which yields interesting sounds. The last example is a modulated version intended to be perceived more natural than the other ones. The MATLAB code of this experiment can be downloaded here.