Plucked Guitar Transients: Comparison of Measurements and Synthesis

The predictions of a calibrated synthesis model for the coupled string/body vibration of a guitar are compared with measurements. A reasonably good match is demonstrated at low frequencies, except that some individual decay rates of “string modes” are hard to match correctly. It is shown to be essential to use an accurate damping model for the string: the intrinsic damping of the classical guitar strings used here is found to vary by an order of magnitude in the audible frequency range. At higher frequencies, two phenomena not included in the synthesis model are found to produce disparities between measurement and synthesis. First, the two polarisations of string motion show modal frequencies which are split further apart than can be explained by body coupling. This effect is tentatively attributed to slight rolling on the fret and/or bridge, resulting in different effective lengths for the two polarisations. Second, unexpected extra peaks are seen in the measured spectra, which are frequencydoubled relatives of the expected peaks due to transverse string resonances. The non-linear mechanism creating these additional peaks is presumed to involve excitation of longitudinal string motion by the transverse vibration.