Analyzing Timbres of Various Musical Instruments Using FFT and Spectral Analysis

Abstract

A sound’s unique timbre is based on the various harmonic frequencies present within its waveform. Through Fast Fourier Transform software, waveforms can be easily decomposed into their component frequencies and a spectral analysis of frequency can be conducted as a method of quantitatively describing timbral characteristics of a sound. In this investigation, the range of frequencies present in a spectrum as well as the average intensity of the first 10 overtones in a sound will be used to classify the timbres of various instruments relative to one another. This will be done by generating a Range-Intensity graph of harmonic frequencies present in sound samples of each instrument.

The results of this investigation reveal that it is not only possible to quantitatively analyze instrumental timbre by generating and mapping out the harmonic frequency data of a specific sound, but that such a quantitative analysis is also incredibly useful. Unlike the traditional, qualitative method of describing timbre, a quantitative analysis would allow for timbral qualities to be transformed into information that can be understood by computers. Today, timbral classification and the decomposition of waveforms has many applications in science and sound engineering. By refining methods for quantitative timbral analysis, it becomes possible to further enhance timbre recognition software and apply such methods to a wider range of technological developments.