Estimates of transmission characteristics related to perception of bone-conducted speech using real utterances and transcutaneous vibration on larynx

Abstract

Beacause transmission characteristics of bone-conducted (BC) speech from the larynx to auditory systems have not yet been clarified, this paper investigates the transmission characteristics related to the BC speech perception focusing on temporal bone (TB) vibration signals and ear canal (EC) radiated speech signals. First, long-term average spectra (LTAS) of the normally produced speech signals recorded at the lips, TB and EC were analyzed. It was found that the frequency components above 2 kHz were relatively decreased in the TB vibration and those below 1 kHz and above 3 kHz were relatively decreased in the EC-radiated speech. Second, transfer functions from the larynx to the observation positions (the lip, TB, and EC) were measured using transcutaneous excitation at the larynx. It was found that the larynx-to-TB transfer functions partially emphasized the frequency region below 1 kHz and the larynx-to-EC transfer functions attenuated the frequency components below 1 kHz and above 3 kHz. These results indicate that the lower frequency components of BC speech are transmitted through TB vibration and the higher frequency components are transmitted through EC-radiated speech.