Machine Learning Methods for Estimating the Excitation Point on a Plucked String of a Classical Guitar

Abstract

The classical guitar is described as a "miniature orchestra" due to the various tone colors that it can produce. One parameter that can control the timbre on the guitar is the excitation point on the string. In this study, a machine learning model was built to determine the excitation point on the string given an audio signal. It was noted that including the mel-frequency cepstral coefficients in the feature set yields outstanding performance. Furthermore, principal component analysis can be used to reduce the dimensions without sacrificing much on performance. Among three well-known algorithms, it was observed that the multi-layer perceptron yields the best performance in terms of classification and regression. Lastly, the models were trained and tested on different subjects and it was noted that each subject has a unique model of its own since different subjects have different physiological parameters that can affect the produced guitar tone.