This paper proposes a method to measure the impulse response of a system with long reverberation using a rectangular pulse as the source signal. The progress of digital device has provided us with various useful digital equipments such as a howling-cancel system. To realize the system, it is necessary to estimate the impulse response. Aresponse to a pulsive signal has been used, for example, to estimate the reverberation time. However, since the pulsive signal is not equal to the impulse, cross-spectral technique is necessary to estimate the impulse response, and much computation time is required to compute many discrete Fourier transforms. On the other hand, we propose a method using a rectangular pulse with a finite amplitude. Under some conditions developedin this paper, the response of the system to the rectangular pulse can be approximately equal to the impulse response. Therefore, it is not necessary to compute Fourier transforms, and the method requires the least amount of computations. We describe first the conditions to estimate the impulse response accurately using a rectangular pulse. Next, an index is introduced to evaluate the accuracy of the estimated impulse response. Using this index, the conditions are investigated in detail by computer simulations. Actual room experiments are also carried out to show that the evaluation method by the index is actually effective in practical cases, where there are external noises.

View full abstract

A covariance method is proposed to analyze speech more accurately than an autocorrelation method. The new covariance method, called the non-Toeplitz form method, ensures uniqueness and stability of data analysis. First, forward and backward prediction models are presented. The uniqueness of definitions of a data vector and a prediction error vector is proved. Second, prediction procedures based on Gram-Schmidt orthogonalization are discussed, and new sample autocorrelation functions are defined. Third, prediction procedures for solving normal equations by using matrix decomposition are presented. An equivalence relationship between auxiliary variables in this procedure and PARCOR coefficients is also discussed. Fourth, the positive definiteness of an autocorrelation matrix whose elements are given by the autocorrelation function proposed in this paper is shown. Finally, the proposed method is compared with other methods.

View full abstract

Particle velocity of the system of a pair of square transducers and that of circular ones are computed precisely and are compared with each other. The principle for the computation is based on Rayleigh's theory, and numerical multiple integration is achieved. In the computation, repeated trapezoidal rule is used, and how the result converges with the increase of the number of subintervals is investigated. The convergency of the system of a pair of square transducers by numerical quadruple integration and that of circular transducers by numerical double integration with analytical decrease of integral multiplicity are revealed to be different. Particle velocity and velocity potential multiplied by 2 π/λ show almost the same convergency.

View full abstract

The present paper describes the effect of a time window on the accuracy of impulse response estimation using the cross-spectral technique. Time windows used in this technique have been investigated both for time discrimination and for frequency characteristics (i.e., band width of the main lobe and magnitude of the side lobes). However, such investigations have not fully accounted for why the accuracy of estimated impulse responses is decreased. This paper describes those reasons, and recommends what type of time window should be used to accurately estimate the impulse response for two typical cases: one, where the transfer function is expressed as a simple time delay; and the other, where the impulse response is represented by an exponentially decaying white noise-the most simple model of an impulse response of a room.

View full abstract

Cross-cultural surveys on neighborhood noise problems were conducted in Japan, Germany and England in 1980 and 1983. The main findings are as follows:(1) Some differences were found between Japan and Germany in sounds which residents were aware of or annoyed by.(2) In both countries the use of loudspeakers was accepted for conveying information necessary to the respondents.(3) German respondents seemed to find it more difficult to become habituated to noise, and to be less tolerant of being annoyed by noise from neighbors.(4) When annoyed by noise from neighbors, German respondents tended to take more direct actions. They also chose more direct countermeasures against noise.(5) In the opinions concerning neighborhood noise problems, Japanese. respondents had more critical attitudes.(6) The affective meanings of some terms related to noise (e.g. “loudness” and “noise”) were different in different languages. The differences found in these surveys seem to derive from the cultural backgrounds of the countries.

View full abstract

In this report, when a general stationary random noise of non-Gaussian arbitrary distribution type is attenuated by various kinds of sound insulation systems, such as a singlewall, a double-wall and a barrier, a simplified probabilistic evaluation method for the output random noise fluctuation is theoretically proposed in a unified form of the probability distribution expression. The validity of the proposed evaluation method is experimentally confirmed by applying it to the actual noise data.

View full abstract