Project "Acoustic coin machine". In industrial practice the acoustic pattern recognition is often used, apart from speech recognition, for example, in the manufacture of glasses, tiles and roofing tiles. Items that are defective as a result of cracks or cavities are rejected. Thus, a defective glass sounds different from a perfect glass..
(a) Think constructively about the mechanical structure of such a device. After a certain distance travelled (drop height, for example, 40 - 50 cm) the coin drops on to a stone or marble top. The impact noise (tone!) must be recorded using a microphone. On the basis of the acoustic pattern the denomination of the coin can be recognized (2 €, 1 €, 50 c, 20 c, 10 c)! (b) Why should only pattern recognition be successful in the frequency range?
(c) Record appropriate test series for all coins. Compare the spectra of the different coins. Why are there noticeably so many identical frequency values in the spectra of all coins?
(d) Now make appropriate use of the parameter module to filter out the different parts of the coin spectra.
(e) Create the neuronal network analogous to "speech recognition"