Traffic noise penetration into a built-up area is investigated using the concept of the A-weighted sound exposure level. Geometrical spreading of the noise and the ground effect have been taken into account. The theory is based on the premise that buildings affect the propagation much more through reduction in the direct waves than through enhancement by diffraction and reflection.

View full abstract

The multi-layer perceptron (MLP) trained by a back-propagation (BP) algorithm pro ducesa posterior probability. However, the MLP cannot reject input data of thecategories that the training data set does not include. On the other hand, if the threshold is defined, template-matching can reject the input data when the distance between input data and each template exceeds the threshold. In this paper, we present a template matchingneural network (TMNN), which has the characteristic of template-matching, and a Sub-Branch (SB) training algorithm. In order to confirm the effectiveness of the TMNN on phoneme recognition, we compared the result with the MLP, training by the BP algorithm.

View full abstract

In this paper, a new trial of stochastic evaluation problem on the output sound level fluctuation of a sound insulation system with Gaussian random wave excitation is pro posedby considering the internal mechanism based on a frequency transfer character of insulation system and a mean-squaring operation of the sound level meter. The charac teristicfunction for the response probability of the sound level meter is derived first by employing the Parseval's complete relation. More concretely, an explicit expression of the output level probability distribution is theoretically given in a general expansion form by applying a well-known residue theorem in the theory of functions. Then, the output response probability function of the sound insulation system can be predicted by using the frequency spectrum of input random noise, the frequency transfer character of the noise insulation system and the measurement mechanism. Finally, the theoretically predicted results are compared with the experimentally observed results using a usual sound level meter for two kinds of single wall inserted between two reberveration rooms.

View full abstract

The results of experiments examining the localized direction of a tone in the presence of band limited pink noise (300-8, 000Hz) are described. Both real and virtual sound images have been considered. It has been found that for signal frequencies of approxi mately1kHz or less the perceived image position is shifted away from the noise source regardless of image type. For higher signal frequencies, the sound localization for a tone was markedly disrupted by the presence of interfering noise. The shift of localized direction of a signal away from the noise may be explained as being due to a kind of spatial masking between them. This would cause a deformation of the sound image of the signal which in turn results in the biasing of a ‘perceived center’ of the signal sound image away from the noise.

View full abstract

In our daily lives we are surrounded with a variety of sounds which vary with time and spread in space. The configurations of sound along the temporal stream have the meaning. The meaning will be destroyed if these sounds are divided into short seg mentsas used in conventional methods. The method of continuous judgment by category and the method of continuous judgment by selected description have been developed for examining the impression of temporally varying sounds along the temporal stream. In this paper these methods are introduced with their procedures and applica tionsto the evaluation of helicopter noise.

View full abstract

This paper presents an unsupervised speaker adaptation method from short utterances. The code spectra for the templates are adapted to those of an input speaker by interpolating the estimated speaker-difference vectors for given typical spectra. These dif ferencevectors are estimated so as to minimize the fuzzy objective function for the adapt edreference codebook under some constraints. The fuzziness (F) and constraint pa rametersare examined using SPLIT-based word recognition tests with 28-word vocabu laryand reference patterns from a male speaker. Using 1.8 s training samples, the results show that the proposed method with F=1.5 gives a 9.0% higher recognition rate for the four male speakers than the minimum VQ distortion method (F=1.0). For 20 male speakers, this method improves the average recognition rate from 92.5% without adaptation to 97.5% using 3.6s training samples. Furthermore, a sequential adaptation scheme attains an average recognition rate of 97.4% using test speech itself for adapta tion.

View full abstract