The stationary sound level was measured as the point of subjective equality (PSE) to randomly fluctuating sound by loudness discrimination test. The fluctuation was specified with two parameters, fluctuating width and velocity, which were obtained from the distribution of envelope of sound. The stimuli presentation was carried out by a kind of stochastic approximation in which the level of stationary sound was varied sequentially by observing the response, thereby PSE was roughly estimated. Furthermore, final PSE was presumed in a maximum likelihood estimation by numerical technique. The result is that the sound is louder as the fluctuating width is larger and velocity is faster. To explain the result, a nonlinear temporal integration model of hearing system is proposed. The distribution of model's output has several frequently occurring levels. The average of those levels is closely related with PSE from the experiment.

No abstract

Empirically, 1 s is widely used as the sampling period in the digital measurement of sound level. In order to clarify the theoretical background of this optimum sampling period, we consider the correlation between the data of each sample and discuss the effectiveness of the data from statistical point of view. In the case of general road sound, the optimum of sampling period is the value of Δt for which the autocorrelation coefficient ρ(Δt) of the wave form is equal to 12, where the envelope wave form is rms value obtained by the integration circuit corresponding to the FAST dynamic characteristics, by weighting the sound with A characteristics. This value is a function of the distance between the lane and the measuring point on the road and the vehicle speed. We also measured various kinds of road sounds and obtained the autocorrelation of the time series for each kind. For present status of the road condition in Japan, the optimum of sampling period is in the order 0.5–1.0 s.

To evaluate the expiratory sounds produced during swallowing recorded simultaneously with videofluorographic examination of swallowing (VF) using fast Fourier transform (FFT), and to examine the relationship between dysphagia and its acoustic characteristics. A total of 348 samples of expiratory sounds were collected from 61 patients with dysphagia whose expiratory sounds were recorded during VF. The VF results were evaluated by one dentist and categorized into three groups: safe group (SG), penetration group (PG), and aspiration group (AG). The duration and maximum amplitude of expiratory sounds produced were measured as the domain characteristics on the time waveform of these sounds and compared among the groups. Time window-length appropriate for FFT and acoustic discriminate values (AD values) of SG, PG, and AG were also investigated. The groups were analyzed using analysis of variance and Scheffé's multiple comparison method. The maximum amplitude of SG was significantly smaller than those of PG and AG. The mean duration in SG (2.05 s) was significantly longer than those in PG (0.84 s) and AG (0.96 s). The AD value in SG was significantly lower than those in PG and AG. AD value detects penetration or aspiration, and can be useful in screening for dysphagia.

No abstract