Perceptual Estimation of the Bone Conduction Path for One’s Own Voice
* Presenting author
Abstract:
Reproductions of one’s own voice often sound unnatural to the listener due to the missing bone conduction component. The transfer function describing that component has been previously characterized by exploiting masked thresholds [Pörschmann (2000), Acta Acustica 86]. However, this approach requires specialized sound reproduction equipment to eliminate the air conducted sound path without introducing occlusion effects.In our study, we investigate whether optimizing perceptual similarity between filtered recordings of the participant’s own voice and their perception while speaking yields the same transfer functions as presented by Pörschmann. To do this, we first add an estimated bone conduction component to the recordings and then optimize this component perceptually.In this contribution, we present the experimental paradigms used to achieve perceptual similarity. We discuss preliminary findings and outline how that knowledge could be applied in psychoacoustic studies with own-voice stimuli.