Measuring the haptic behavior of an acoustic guitar as perceived by the player by means of a vibrating actuator

Abstract

Two sets of recordings of the vibration produced by plucking the fifth and the second string of an acoustic guitar were acquired using an accelerometer secured to the neck of the instrument. Vibrations from both sets could be reproduced using a recoil-type vibrating actuator attached at the neck of the guitar. In one of the sets, salient spectral features of the original recordings were altered. We performed a preliminary study involving nine volunteer participants, blindfolded and artificially deafened using earplugs and loud white noise played through headphones. They were asked to discriminate, by holding the neck of the instrument with their left hand, between “fake” (i.e., actuator-produced) or real vibrations, produced by the experimenter plucking either string two or five. Our aim was to assess if any of the spectral features altered in the second set of recordings increased the recognition rate of actuator-produced vibrations as being “fake.” These features, if present, would then be likely to carry crucial information and should be therefore modeled with extreme care in the simulation of the haptic behavior of the instrument. Results show that, at least for string five, we were capable of identifying one feature (a peak in the vibration spectrum located at 548 Hz), which, if altered, made the recognition rate as “fake” rise, statistically significantly, from 55% to 89%.