In this paper, we propose a new method for evaluating the separation and dereverberation performance of a convolutive blind source separation (BSS) system, and investigate a separating system obtained by employing frequency domain BSS based on independent component analysis (ICA). As a result, we reveal the acoustical characteristics of the frequency domain BSS for convolutive mixture of speech signals. We show that the separating system removes the direct sound of a jammer signal even when the frame length is relatively short, and it also reduces the reverberation of the jammer according to the frame length. We also confirm that the reverberation of the target is not reduced. Moreover, we propose a technique, suggested by the experimental results, for improving the quality of the separated signals by removing pre-echo noise.
Psychoacoustic experiments were conducted to investigate absolute cues for auditory distance in an anechoic room. The presence of absolute cues allows a listener to judge the distance of a sound image with little or no prior experience regarding stimuli or environment. Therefore, the first presentations of the stimuli to the listeners were analyzed. In the first experiment, listeners were divided into three groups based on the distance from a frontal loudspeaker (0.5 m, 1 m, or 4 m) and each group was also divided into four subgroups based on the type of stimulus used in the first presentation. No significant difference was found in the loudspeaker distance or in the type of stimulus by the analysis excluding non-frontal perceptions. The second experiment, in which the loudspeaker was situated at 60 degrees azimuth, demonstrated that the perceived distances of the first presentation were significantly different between the loudspeaker distance of 0.5 m and the loudspeaker distances of 1 m and 4 m. These results imply that the strength of absolute cues in the lateral direction is stronger compared to the frontal direction. Considering the results of measurements on distance-dependent head-related transfer functions reported in previous studies, interaural level differences seem to provide an absolute cue in the lateral direction of 0.5 m.
A new method of reducing the jet screech is explained. A spherical reflector is placed at the nozzle exit for reducing this screech component. The placement of the spherical reflector at the nozzle exit controls the location of the image source as well as minimizes the sound pressure at the nozzle exit. The weak sound pressure does not excite the unstable disturbance at the exit, thus the loop of the feedback mechanism cannot be accomplished and the jet screech is not generated. The technique of screech reduction with the flat plate is also examined and compared with the present method. A good and effective performance in canceling the screech component by the new method is found by the investigation. Experimental results indicate that the new system suppresses not only the screech tones but also the broadband noise components and reduces the overall noise. The spherical reflector is very effective in reducing overall sound pressure level in the upstream region of the nozzle compared to a flat plate. The proposed spherical reflector can, accordingly, protect the upstream noise propagation.
A method established by ICBEN (International Commission on the Biological Effects of Noise) Team 6 (Community Response to Noise) has been used to develop equivalent noise annoyance scales in nine languages. A key assumption of this method is that subjects who speak different languages interpret the concept of “highest degree” of annoyance similarly. In this study, an experiment was conducted using 73 bilingual subjects to test this assumption and thereby assess the equivalence of the Japanese and English ICBEN scales. All of the bilingual subjects followed a slightly modified version of the ICBEN method to produce scales in both Japanese and English. The results indicate that English- and Japanese-speaking subjects do not differ significantly in their interpretations of the “highest degree” of annoyance. Thus, a key premise of the ICBEN method for producing equivalent scales in different languages was confirmed for English and Japanese. In addition, although the Japanese modifier “hijoni” has frequently been translated as “very,” the results of this study show that “extremely” is a more appropriate translation. Finally, evidence was found that bilingual subjects may differ significantly from monolingual subjects in their interpretations of certain words.
When a gas-filled closed tube is driven near the resonant frequency of the gas-column, large amplitude standing wave is induced in the tube. The use of large amplitude standing wave in practical field is based on the exploitation of high amplitude pressure fluctuation together with the associated phenomena like acoustic streaming, thermoacoustic effect and mean pressure distribution. It is already known that shock wave appears in cylindrical tube and thereby the wave motion is limited in the peak pressure that can be generated. In shaped tubes, it is possible to avoid the problem of shock formation and as a result, large amplitude standing wave can be achieved. And also the mode of standing wave induced in shaped tubes depends on the tube geometry. For practical application of large amplitude standing wave, the selection of optimum tube shape must be required. In this report, experimental results are described for three types of tube geometry, such as conical, exponential and half cosine. HFC134a was selected as the working gas for the reason that the effects of tube geometry in the wide range of pressure amplitude can be investigated than air. The frequency responses and pressure fluctuation at closed end of each tube were monitored. The result revealed that the half cosine geometry is suitable in the case of design of the acoustic compressor.