We explore the phenomenon of amplitude variation of bone-conducted (BC) speech compared with that of air-conducted (AC) speech. During vocalization, in addition to the AC components emitted through the mouth, vibrations travel through the vocal tract wall and the skull bone before they arrive at the cochlea. A bone-conductive microphone placed on the talker's head can partly capture these vibrations and convert them to BC speech signals. The amplitude of this BC speech is influenced by the mechanical properties of the bone-conduction pathways. This influence is related to the vocal tract shape, which determines the resonances of the vocal tract filter. Referring to these resonances as formants of AC speech, we can describe the amplitude variation of BC speech with respect to the location of the formants of AC speech. In this work, the amplitude variation of BC speech of Japanese vowels, CV (consonant–vowel) syllables, and long utterances have been investigated in terms of the locations of the first two formants of AC speech. Our observation suggests that when the first formant is very low with a higher second formant, the relative amplitude of BC speech is amplified. On the other hand, a relatively high first formant and lower second formant of AC speech cause a reduction in the relative BC amplitude.

Blind extraction of convolutive speech mixtures can be achieved by the approximate joint diagonalization (AJD) approach. In this paper, we present a least-squares AJD (LS-AJD) algorithm, called fast diagonalization, implemented by minimizing the direct and indirect LS criteria (FDMDI) algorithm. The proposed approach is based on an alternate minimization of the indirect and direct least-squares criteria to the diagonal matrices in the first phase and to the mixing matrix in the second phase, respectively. In our proposed approach, the constrained LS-AJD estimation problem is solved by the method of Lagrange multipliers; moreover, the mixing matrix is estimated by a noniterative method without a nested loop in the second phase. The simulation result demonstrates that overdetermined FDMDI blind source extraction (BSE) provides more effective extracted signals than determined FDMDI BSE in an actual acoustic environment.

We simulated acoustic interaction between the voice source system in the larynx and the acoustic filter of the vocal tract. The vocal tract of a soprano was first scanned in three dimensions using magnetic resonance imaging while she produced four musical notes (A3, E4, D5, and A5) with /a/ and /i/. These images were used to simulate voice production, including the vibratory motion of the vocal folds and the behavior of glottal airflow. Images for the /i/ vowel were used in the simulation, because a good proximity relationship was found between the fundamental frequency and the first impedance peak of the vocal tract. The simulation results revealed that the fundamental frequency (vibration frequency of the vocal folds) was decreased to a large extent by the interaction especially when their natural frequency was in the proximity of the impedance peak. In a specific case, the acoustic load of the vocal tract exerted on the vocal folds changed as a result of the interaction, so the vibratory motion was effectively assisted. These interaction effects were also examined in terms of the phase relation among the temporal waveforms of the glottal variables.

This paper presents an expanded theory for relating airborne-sound-excited and force-excited sound radiations from solid structures. Although the reduction of these two types of sound radiation is a fundamental issue in noise-control engineering, each of them has been historically treated as a separate issue. The reduction in the former is normally called airborne sound insulation and separated from the latter, especially in architectural acoustics. A previous study (M. Yairi et al., J. Acoust. Soc. Am., 140, 453–460 (2016)) established a fundamental relationship between the sound radiations from random-incidence sound-excited and point-force-excited vibrations of a single-leaf infinite elastic plate. A conversion function that relates the two excitation cases was presented in a simple closed form, not including the parameters of the plates, which included of only the specific impedance and the acoustic wavenumber of the medium surrounding the plate. In this paper, the applicability of the conversion function is expanded from a single-leaf infinite elastic plate model to a double-leaf infinite elastic plate model. The sound radiation from a double-leaf infinite elastic plate driven by random-incidence sound and that driven by point-force excitation are theoretically investigated. The conversion function derived from the present model successfully relates the two excitation problems at all analyzed frequencies and has been shown to agree with the previously established single-leaf theory.

In this study, a two-dimensional (2D) mechanical model of the cochlea, discretized by a nonuniform grid, is applied to investigate the mechanisms that limit the increase in the computational efficiency. To amount for experimental findings, cochlear models have become complicated. A cochlear model consists of micro- and macro mechanical models. Many types of micro mechanical model have been proposed. However, macro mechanical models are described by the Laplace equation and show various patterns of the cochlear response depending on the location. Therefore, an efficient step width depends on the location in the cochlea. To resolve this issue, a numerical calculation has been applied to divide the space of a cochlear model into a nonuniform grid and to achieve improved efficiency of the model. However, the limitation of this method remains unclear. To investigate this point, we develop a state space model for 2D cochlear mechanics with a nonuniform gird. Stability analysis and simulations are conducted for the cochlear model with nonuniform and uniform grids. As a result, the number of segments is reduced by 29%. In addition, the execution time is reduced by 10-fold. Therefore, it is shown that a nonuniform grid can efficiently divide the space for cochlear modeling.

A vibroacoustic numerical method employing a finite-difference time-domain (FDTD) method, in which the target floor structure consisting of a floor panel supported by support legs on a floor slab is modeled as a composition of two-dimensional plate elements for the double plate structure and one-dimensional bar elements for the support legs, is proposed. While floor impact sound is difficult to accurately predict owing to the complexity of the vibroacoustic mechanism influenced by the coupling phenomena with the vibration of the double-plate structure connected by support legs and the sandwiched air layer between the double plates. In this paper, the basic theory of the proposed numerical scheme was validated through comparison with excitation test on an acrylic scale model, and the applicability of the method to a practical case of a vibroacoustic transmission via a two-layered floor structure was discussed through a numerical case study.