To clarify human speech perception and production mechanisms, perceptual properties of speakers' own voices transmitted via bone conduction should be further understood. In our previous studies, transfer functions of ear-canal sound pressure and regio temporalis vibration relative to oral-cavity sound pressure were physically measured. The former and latter were assumed to represent perceptual properties of the outer-ear and middle/inner-ear part of bone-conducted (BC) speech transmission during voicing, respectively. This paper investigated the contribution of BC speech reaching the middle/inner ear to one's own voice perception, as well as that reaching the outer ear, by making two types of filtered speech signals based on the measured transfer functions. The mixing ratio of speech signals filtered by each transfer function (i.e., mimicked BC speech signals) relative to the original air-conducted (AC) speech signals were determined through subjective evaluations. It was found that two filtered speech signals were mixed at almost the same level, and a combination of two filtered speech signals were mixed at almost the same level as the AC speech. These findings suggest that the middle/inner-ear part of transmission contributes to BC speech perception at almost the same extent as the outer-ear part.