Timbre de um instrumento musical

The representation of the timbre of a musical instrument involves problems of great complexity. Despite the already known correlation between timbre and the spectral content of the sound, the mapping the spectral characteristics of the great variety of sounds produced by an instrument demands specific methodologies of data analysis. This study presents an approach for this problem through the mapping of the amplitude time-varying curves of the harmonic components, extracted through the Discret Fourier Transform, using techniques of Analysis of Principal Components (PCA). The orthogonal bases defined by the PCA made possible a significant data reduction and the creation of timbral sub-spaces capable of representing the sounds of the instrument in several pitch and intensity levels. Techniques of data classification allowed a semantic analysis of these timbral spaces, making possible the classification of groups of similar timbres in this space and ratifying PCA as an efficient tool for representing the dynamics of the timbre of musical instruments.