The transfer function and the impulse response of an acoustical barrier in front of a large building facade were calculated. The sound field prediction around the barrier was based on the Kirchhoff-Rubinowicz theory, where the velocity potential on the back of a barrier was treated as zero, and then its inverse Fourier transform was calculated as the impulse response. Multiple reflections, in the back of a barrier, the building facade and the ground, were calculated only for specular reflections of diffracted waves which were born at the edges of the barrier. For damping of these multiple reflections, absorbingmaterial was lined on the back of the barrier and the ground. These calculation methods were compared with the scale model experiments in the time domain giving good agreements. The diffraction in the back of a barrier became more intuitive being expressedin the time domain.

抄録全体を表示

Accentual rules are constructed for the phonological component of a semantics-to-speech system of Japanese in a quest for possible integration of management of segmental and accentual phonological rules that bear an essential part of the intrinsic knowledge for speech synthesis from concept. Six accent assignment and two accent movement rules are counted, with the former placed immediately after the segmental rules in the same domain of cyclic rule application and the latter applied postcyclically between the noncyclicsegmental and phonetic rules. Segmental and accentual phases of rule application are found to be successfully integrated by using the principle of recursive agglutination of a suffix to a stem with the former carrying morphophonological and accentual conditionsfor the latter to meet.

抄録全体を表示

A new speech recognition system using the neurally-inspired Learning Vector Quantization (LVQ) to train HMM codebooks is described. Both LVQ and HMMs are stochasticalgorithms holding considerable promise for speech recognition. In particular, LVQ is a vector quantizer with very powerful classification ability. HMMs, on the other hand, have the advantage that phone models can easily be concatenated to produce long utterance models, such as word or sentence models. The new algorithm described here combines the advantages inherent in each of these two algorithms. Instead of using a conventional, K-means generated codebook in the HMMs, the new system uses LVQ to adapt the codebook reference vectors so as to minimize the number of errors these referencevectors make when used for nearest neighbor classification of training vectors. The LVQ codebook can then provide the HMMs with high classification power at the phonemiclevel. As the results of phoneme recognition experiments using a large vocabularydatabase of 5, 240 common Japanese words uttered in isolation by a male speaker, it was confirmed that the high discriminant ability of LVQ could be integrated into an HMM architecture easily extendible to longer utterance models, such as word or sentencemodels.

抄録全体を表示

A new method for measuring a delay time is proposed. If an observed signal is regardedas an output signal of a minimum phase system with delay, its delay time can be estimated by use of the Hilbert transform relationship between the log magnitude and phase of the transfer function of the minimum phase system. Discussions are mainly focused on practical cases where the observed signal is band limited. A procedure for extrapolating the frequency characteristics of the signal is also presented to minimize the estimation error due to the bandwidth limitation. The method is applied to measure the boundary location of acoustic materials whose reflection characteristics are minimum phase. The estimation error less than ±0.7 mm in distance is achieved when the test signal of 5 kHz bandwidth is used. As an example, the location of the reflecting surface of a gravel layer is estimated. Using this result, the acoustic impedance of gravel is computed.

抄録全体を表示

Measurements on acoustic noise emission from the downwind 15m diameter horizontal axis wind turbine generator (installed at Mechanical Engineering Laboratory under MITI's Sunshine project) have been carried out for the purpose to obtain basic data necessary for a future noise reduction and assessment study. In this paper, characteristicsof aerodynamic and mechanical noise of the machine in an audible frequency range are presented and a mechanism of noise sources is discussed. Further, the reference sound level and the directivity index are determined according fundamentally to the IEA Recommended Practice on acoustic noise measurements. Also the dependence of noise on wind speed and operating conditions is presented.

抄録全体を表示