Tanetoshi Miura

Spectral Properties and Quantitative Evaluation of Hypernasality in Vowels

A new technique for evaluating hypernasality using an acoustic approach is presented. In a preliminary study using this technique, nasal resonance was assessed in 17 normal subjects and 16 subjects judged to be hypernasal. Analyses of the one-third-octave power spectra revealed an increase in power level between the first and second formant, and a reduction in the power level in second and third formant regions among utterances judged to be hypernasal. Factor analysis of the perceptual ratings revealed that the consensus perception of hypernasality accounted for 71% of the total variance. An additional 8% was accounted for by individual differences. Multiple regression analysis revealed a high correlation between the consensus perception of hypernasality and the variance in two acoustic-power levels, these being the power level between the first and second formant and the power level of the second and third formant regions.

An adaptive noise control system in air‐conditioning ducts

This paper discusses an adaptive control system for the active cancellation of acoustic noise in air‐conditioning ducts. Recently, in several systems for active noise control, efforts have been focused on the use of adaptive digital filters to implement the system controller used to produce the artificial sound. The approach used in the newly proposed model assists in canceling broadband noise and also helps in improving the adaptation speed for colored noise. This model consists of a system identification (IDT) process and an adaptive noise cancellation (ANC) process. Two stand‐alone types of adaptive controllers using the fast‐least‐mean‐squares (FLMS) algorithm and the variable‐step least‐mean‐squares (VS‐LMS) algorithm were used in various experiments. Results of experiments in actual air‐conditioning ducts are presented.

Prefiltering method for a head‐related stereophonic system

The prefiltering scheme of the usual head‐related stereophonic system using two loudspeakers for reproduction [M. R. Schroeder et al., J. Acoust. Soc. Am. 56, 1195 (1974)] was modified. In addition to the usual filtering scheme, an additional filtering stage is introduced that converted the frontal incident characteristics of the dummy head into those for individual listeners. Using this filtering scheme, the exact reproduction of the frontal sound image becomes possible, which is difficult when the sound recorded through a dummy head is played back. The ability of this modified system to reproduce the original sound localization by a listening test in an anechoic chamber is evaluated. In the test, both the conventional scheme and the new one were evaluated. As a result, the localization of the frontal sound image was reproduced exactly using the new system, and the conventional system failed in this reproduction. It was also confirmed that sound from other directions in the three‐dimensional space was al...

An electronic sound cancellation system for air‐conditioning ducts

Recently, research on an electronic sound cancellation system that is able to cancel noise using sound of the opposite phase has been conducted. This kind of noise control technique is referred to as active noise control (ANC). In order to realize a practical working ANC system, it has been considered desirable to apply the LMS (least‐mean‐square) algorithm to such a systems control, especially the VS (variable step)‐LMS algorithm due to its high speed of convergence. The realization of an ANC system of the stand‐alone type is presented. Experiments were conducted in actual air‐conditioning ducts having a cross section of 500 × 500 mm by using the realized stand‐alone type active noise control system with the VS‐LMS algorithm. Below the cross‐mode frequency, more than 15 dB of attenuation was obtained for broadband random noise, even in the presence of an air flow of 3.7 to 9.0 m/s.Recently, research on an electronic sound cancellation system that is able to cancel noise using sound of the opposite phase has been conducted. This kind of noise control technique is referred to as active noise control (ANC). In order to realize a practical working ANC system, it has been considered desirable to apply the LMS (least‐mean‐square) algorithm to such a systems control, especially the VS (variable step)‐LMS algorithm due to its high speed of convergence. The realization of an ANC system of the stand‐alone type is presented. Experiments were conducted in actual air‐conditioning ducts having a cross section of 500 × 500 mm by using the realized stand‐alone type active noise control system with the VS‐LMS algorithm. Below the cross‐mode frequency, more than 15 dB of attenuation was obtained for broadband random noise, even in the presence of an air flow of 3.7 to 9.0 m/s.

Relations between frequency selectivity, temporal resolution, and speech intelligibility in sensorineural hearing‐impaired subjects

Frequency selectivity (the shape of the auditory filter), temporal resolution (the ability of gap detection), and speech intelligibility were measured for hearing‐impaired subjects, and the relationships among these were investigated. The subjects were aged 18–49 years and had sensorineural hearing loss. Their average hearing threshold level was above 70 dB HTL. The shape of the auditory filter was measured from pure‐tone threshold values under various maskers of notched broadband noise. Gap detection thresholds were measured using octave‐band noise as a stimulus. A speech intelligibility test was conducted using two‐syllable female speech. The results showed no correlation between hearing threshold level and any one of the following three parameters: auditory filter shape, gap detection threshold, and speech intelligibility score. However, the present subjects with severe hearing loss also showed a strong correlation between the sharpness of auditory filter and the speech intelligibility score, and a wea...

Sound reproduction in a car cabin using a digital filter network

Designs to reproduce stereo sound fields in a car cabin have many problems, such as the asymmetrical positions of loudspeakers (for the listener) and the complex characteristics of the sound transmission. Therefore, a technique is investigated in which the sound field in a car cabin is transformed using a digital filter network, keeping in mind the interaural sound pressure in both ears of the listener. The digital filter network can be realized from the transfer functions measured by a head and torso simulator in the car cabin and in the desired sound field. The purpose of using the digital filter network is to equalize the sound pressure in both ears of the listener (when the digital network is not used) to that obtained in the desired sound field. The realizability of the desired sound field in a car cabin was confirmed by computer simulation.

Digital equalizer of a dummy head system using a direct approximation technique of log magnitude response

When we intend the whole‐space stereophnic system, we must design this system to be simulated for original listening conditions. For this purpose, we need several equalizing networks [M. R. Schroeder and D. Gottlob, J. Acoust. Soc. Am. 56, 1195–1201 (1974)]. Since these equalizers have rather complicated responses, we constructed them by digital systems. The digital equalizers described here have a form of digital filter which presents the best mean square approximation. To determine the coefficients of these filters, a direct approximation technique of log magnitude response for digital filter, proposed by I. Imai [IEEE Trans. Acoust., Speech Signal Process. 25, 127–133 (1977)], was utilized. The resulting digital equalizers are recursive type and realized in a cascade connection of very simple second‐order‐type recursive digital filters (number of multipliers is two in this second‐order type filter). These equalizers have several important advantages, i.e., fewer multipliers are needed in these construc...