To predict noise within 100 meters from a bridge and within a few meters above the ground, measurements of the sound exposure level, LAE, at four positions are required. The process of noise generation is modeled by a homogeneous line of point sources with dipole directivity. Noise propagation is governed by geometrical spreading and ground effect, which is modified by the other wave phenomena, particularly atmospheric turbulence. To illustrate this method, measurements and calculation were performed at the Nishitetsu Line Bridge over the Nakagawa River in Fukuoka.

View full abstract

This paper is concerned with the vibrations of a loudspeaker's mechano-acoustical part, a membrane. The dynamic behaviour of loudspeakers, one with a planar honeycomb membrane and one with a quasi-conical paper membrane is presented. An electromechanical driving system of the loudspeaker was not considered. Two experimental conditions were analysed: loudspeakers were either mounted in enclosures (a closed-box type) or they were freely suspended. The structural behaviour of loudspeakers was evaluated by means of an experimental modal analysis technique. A set of Frequency Response Functions (FRFs) was measured in the experiment and then modal parameters (modal frequencies, modal damping and mode shapes) were extracted. Modes of the planar loudspeaker were qualitatively compared with the theoretical modes of a circular plate. The structural behaviour of the quasi-conical loudspeaker was compared to the modes of a frustum of a cone. An influence of an enclosure on modal parameters of loudspeakers is discussed.

View full abstract

A method is proposed for determining representatives of the sound localization transfer function (SLTF) aiming at a compromise between feasibility and localization accuracy. This method is based on principal components analysis (PCA). The magnitudes of SLTF measured at both ears of 56 subjects for 24 targets were modelled by linear combination of six principal components (PC). For each target and ear, the representative was determined to be the SLTF whose weights of PC contributing to the magnitude best approximated the average over subjects. The effectiveness of the representatives was evaluated both statistically and psychoacoustically. The statistical analysis indicates that the individual distribution of the weights of PC is assumed to be a six-dimensional normal distribution that is densest around the average. The discrepancy of the representatives from the average is within 1.5 percent of the individual accumulation. The psychoacoustical evaluation for 28 of the subjects indicates that the increase in the front/back confusion rate in the use of the representatives is significant for frontal targets, but not for other targets. The lateral judgement is robust against SLTFs for all targets tested. However, considerable individual dispersion of the front/back confusion rate hindered from determining the allowance of the discrepancy between individual SLTFs and the representatives for acceptable localization accuracy.

View full abstract

In the speaker recognition, when the cepstral coefficients are calculated from the LPC analysis parameters, the prediction error, or LPC residual signal, is usually ignored. However, there is an evidence that it contains a speaker specific information. The fundamental frequency of the speech signal or the pitch, which is usually extracted from the LPC residual, has been used for speaker recognition purposes, but because of the high intraspeaker variability of the pitch it is also often ignored. This paper describes our approach to integrating the pitch and LPC-residual with the LPC-cepstrum in a Gaussian Mixture Model (GMM) based speaker recognition system. The pitch and/or LPC-residual are considered as an additional features to the main LPC derived cepstral coefficients and are represented as a logarithm of the F0 and as a filter bank mel frequency cepstral (MFCC) vector respectively. The second task of this research was to verify whether the correlation between the different information sources is useful for the speaker recognition task. For the experiments we used the NTT database consisting of high quality speech samples. The speaker recognition system was evaluated in three modes-integrating only pitch or only LPC-residual and integrating both of them. The results showed that adding the pitch gives significant improvement only when the correlation between the pitch and cepstral coefficients is used. Adding only LPC-residual also gives significant improvement, but in contrast to the pitch, using the correlation with the cepstral coefficients does not have big effect. The best results we achieved using both the pitch and LPC-residual and are 98.5% speaker identification rate and 0.21% speaker verification equal error rate compared to 97.0% and 1.07% of the baseline system respectively.

View full abstract

A method is proposed for precisely identifying the multiple signal transmission paths of a multi-input single-output acoustic system by using correlated input signals. Such identification has not been previously achieved by a conventional straightforward generalization of the principle for identifying a single-input single-output acoustic system. To this end, pre-filters are applied to the system inputs so that the relation between inputs and output satisfies-the condition of a unique solution set given by the framework ofa diophantine equation

View full abstract

In our daily life, there are many kinds of noise. Some of them are periodic, such as exhaust noise of an engine and fan noise, and numerous methods have been proposed to control the periodic noise. Most methods need a reference signal, and the reference signal must be measured or known in advance. In actual environment, the frequency of the periodicnoise is not always time invariant. When active noise control systems are implemented, the information on a periodic noise can be obtained by an external sensor like tachometer. Therefore the performance of an active noise control system depends on the accuracy of the external sensor. In this paper, two adaptive algorithms able to control a periodicnoise are proposed. One algorithm estimates fundamental frequency, amplitude and phase of a periodic noise. Another algorithm estimates each frequency comprised in a periodicnoise independently. The performance of these algorithms is evaluated by computer simulations and in the experiment using a duct.

View full abstract

A technique of optical beam deflection by ultrasonic waves was developed in order to measure acoustic nonlinearity parameters B/A in liquids. Finite-amplitude ultrasonic waves of 6.65 MHz propagating in a liquid sample deflect a laser beam of normal incidence. The deflection angle was detected with a photodiode having double elements, yielding a distorted waveform. Spectral analysis of the waveform yielded the ratio of the amplitude of the second-harmonic wave to that of the fundamental wave. The absolute value of the fundamental sound-pressure amplitude was obtained independently from the Raman-Nath parameter measured using the same experimental system under the condition that a laser beam waist is larger than sound wavelength. From these measurements, the nonlinearity parameters were evaluated to be 5.3 for water and 10.3 for ethyl alcohol. These results are in good agreement with established values.

View full abstract

MIDI interface is originally designed for electronic musical instrument but we consider this music-note based coding concept can be extended for general acoustic signal description. We proposed applying MIDI technology to coding of bio-medical auscultation sound signals such as heart sounds used in medical diagnosis for constructing medical audio databases. Furthermore, we have tried to extend our encoding algorithm in order to apply to natural environmental sounds such as bird sounds and even to vocal sounds. Utilizing MIDI multi-track feature we could produce high-quality vocal sounds; namely we could make an electronic musical instrument sing a song. In this report, we describe briefly our proposed multi-track MIDI encoding algorithm based on GHA, and present an example of several coded experiments for vocal sounds using this algorithm.

View full abstract