In general, a speech signal can be measured by a microphone, such as a throat microphone. However, the speech signal measured by a microphone often contains surrounding noise. On the other hand, although a throat microphone is effective for surrounding noise, the speech signal it measures includes body-conducted internal noise. In this study, we propose an improvement method for the sound quality of the speech signal measured by a throat microphone to achieve speech recognition well. The relationship between the original speech signal and the speech measured by the throat microphone is not clear. Therefore, we consider the relationship as a multiplicative and additive model of the original speech signal and noise components with unknown parameters. An algorithm is proposed to simultaneously estimate the original speech signal and the unknown parameters using Bayes’ theorem based on the speech signal measured by the throat microphone. Finally, a speech recognition experiment is conducted to confirm the effectiveness of the proposed algorithm.

Single-channel blind dereverberation aims to remove reverberation from a single-channel reverberant signal without using any prior knowledge. In acoustics, weighted prediction error (WPE), a method mainly used for a multi-channel signal, is often applied for this task. However, it is difficult to achieve well-performed dereverberation for a single-channel signal. In this paper, for better single-channel dereverberation, we propose to simultaneously estimate the source signal and the room impulse response (RIR) instead of only predicting reverberation. By modeling convolution using matrix lifting in the time-frequency domain, we formulate the dereverberation problem as a non-convex optimization problem of recovering a sparse rank-1 matrix. In sparse regularization, we introduce reweighting, enabling the improvement of sparse matrix recovery. The alternating direction method of multipliers (ADMM) with acceleration is applied to approximately solve the optimization problem, resulting in closed form updates. In our experiments, we confirmed that the proposed method outperforms existing methods in several reverberant conditions and is capable of removing both early reflection and late reverberation. MATLAB code of the proposed method is available online (https://doi.org/10.24433/CO.3541617.v1).

Amplitude modulations of an aerodynamic sound by sound source movement or convection is investigated experimentally. The directivities of an aeolian tone sound radiated from a cylinder fixed in a wind tunnel were measured with horizontally aligned microphones outside the flow as a simplified dipole component of aerodynamic sound generated by pressure fluctuations, which is dominant at low Mach number. The results show that the acoustic intensity of radiated sound can increase approximately in inverse proportion to the fourth power of the ratio of the apparent wavelength of the radiate sound to the original wavelength. The amplitude modulation of the sound by convection was measured and then compared with the modulation of the same sound by source movement. The result revealed that an aerodynamic sound leads to a larger amplitude modulation by source movement compared with the modulation measured in a wind tunnel experiment, although the apparent wavelength of sound radiated from the same source in the atmosphere changes by the same proportion.

The vocal membrane, i.e., an extended part of the vocal fold, is present in non-human primates. To understand its function in animal vocalization, Mergell et al. (1999) constructed a mathematical model of the vocal membrane and predicted that the vocal membrane lowers the phonation threshold pressure required to initiate the vocal fold oscillations. The present study constructed a physical model of the vocal membrane based on a micro-computed tomography measurement of a rhesus macaque larynx. Our physical experiment confirmed that the phonation threshold pressure was indeed lowered and, consequently, the vocal efficiency was increased by the vocal membrane.

The widespread use of electric vehicles has altered the perception of automobile sounds in recent years. As automated driving becomes more prevalent, it is assumed that a decrease in cognitive load will make vehicle sounds more noticeable, thereby increasing discomfort while driving. Studies have aimed to reduce the noise disturbance of electric vehicles and evaluate their effects from a physiological perspective. However, no study has examined the attentional mechanisms while driving electric vehicles from a neurophysiological perspective. In this study, we investigated selective attention to vehicle sounds and the effects of changes in vehicle sounds on event-related potentials under driving conditions. Event-related potentials are closely associated with the physical characteristics of stimuli, the cognitive processes of perception, and selective attention. We examined event-related potentials, such as P3, during driving to understand their relationship with selective attention in the driving environment. Our study revealed differences in event-related potentials with subtle changes in vehicle sound pressure levels while driving. Furthermore, we found that differences in attentional engagement may be evoked by sound pressure level changes. These results will help investigate practical approaches to reduce vehicle noise and enhance comfort within car interiors.

The C-C method uses cardioid microphones to measure sound pressure, particle velocity, and sound intensity. We attempt to apply the C-C method to measuring the sound transmission loss of a building component. We obtain results in the one-third octave bands above 315 Hz that are comparable to those obtained by the sound pressure level difference method commonly used. The results reveal that the measurement method using the C-C method validated in this study can be an alternative to other sound transmission loss measurement methods.

Solving the permutation problem is essential for determined blind source separation (BSS). Existing methods, such as independent vector analysis (IVA) and independent low-rank matrix analysis (ILRMA), tackle the permutation problem by modeling the co-occurrence of the frequency components of source signals. One of the remaining challenges in these methods is the block permutation problem, which may cause severe performance degradation. In this paper, we propose a simple and effective technique for solving the block permutation problem. The proposed technique splits the entire frequency bands into several overlapping subbands and sequentially applies BSS methods (e.g., IVA, ILRMA, or any other method) to each subband. Since the splitting reduces the size of the problem, the BSS methods can effectively work in each subband. Then, the permutations among the subbands are aligned by using the separation result in one subband as the initial values for the other subbands. Additionally, we propose SS-IVA and SS-ILRMA by combining subband splitting (SS) with IVA and ILRMA. Experimental results demonstrated that our technique remarkably improves the separation performance without increasing computational cost. In particular, our SS-ILRMA achieved the separation performance comparable to the oracle method (frequency-domain independent component analysis with the ideal permutation solver). Moreover, SS-ILRMA converged faster than conventional IVA and ILRMA.

This study adopts the Taguchi method to determine the optimal robust parametric levels for enhancing the sound absorption performance of 316L stainless steel foam. The factors examined in this study are the average pore size (1.1 mm−4.4 mm), the porosity (48%−61%), and the thickness (10 mm−30 mm). Each factorial contribution is evaluated using an analysis of variance (ANOVA). The analysis indicates that porosity is the most significant factor, accounting for 89.4% of the total variance, while pore size and thickness have a smaller influence. There is no interaction between the factors. An increase in the porosity increases the sound absorption coefficient. The optimal combination of factor levels for sound absorption specifies a pore size of 1.1 mm, a porosity of 61%, and a thickness of 30 mm. The highest sound absorption coefficient of 0.7 is attained at a frequency of 6300 Hz. This study proposes a useful approach combining the Taguchi method with ANOVA designed to accelerate the development of the sound adsorption performance of metal foam.

We propose a method using impulse responses measured at multiple points to evaluate the acoustic reflection of materials. This method reproduces a plane incident wave based on the Rayleigh integral and estimates its reflected angular spectrum using spatial Fourier transforms. The IRs are measured for all combinations of sources and receivers, which are both arranged in planar grid patterns. Numerical simulations with the image source method validated the approach for incident angles of 0–60 degrees. However, as the incident angle becomes almost parallel to the material surface, i.e., at grazing incidence, synthesizing wavefronts with sufficient accuracy becomes more challenging.

Yakisugi, a traditional Japanese exterior material manufactured by charring the surface of a cedar board, has garnered interest overseas, particularly for its application as an interior material. This study investigates yakisugi’s sound absorption characteristics, expecting its porous carbonized layer to absorb sound. Particularly, considering its use as an interior material, this study experimentally investigates the impact of protective surface coatings. Using an impedance tube and a reverberation chamber, the results suggest that although the surface coating slightly decreases the sound absorption coefficient at higher frequencies, yakisugi maintains a certain level of sound absorption.

We proposed an alaryngeal speech system using an acoustic tube and a device for flexible voice control. The system interface enables users to adjust the voice pitch by changing the hand angle and to switch a pulsed signal and white noise as the sources of the voice using three switches. This allows free adjustment of the pitch within a three-octave range and switching vocalization methods with one hand, making it possible to reproduce fricative and plosive sounds such as [s] and [k], even to sing a song.

Previous studies have shown that speech spoken by someone familiar is more intelligible than an unfamiliar voice in the presence of a competing talker. Although the voices of friends or family members were the familiar voices in these previous studies, other studies have shown that one’s own recorded voice is more familiar than others’ voices. However, the intelligibility of one’s own recorded voice still remains unclear. The aim of this study is to investigate the intelligibility of one’s own recorded voice compared with the voices of others. We conducted a large-scale online Japanese version of the coordinate-response-measure (CRM) task including one’s own and others’ recorded voices with 137 Japanese participants. The results showed that one’s own recorded voice is more intelligible than others’ voices when there is a competing talker. We also analyzed the relationship between the performance of the CRM task and impressions of one’s own recorded voice. The results indicated that the intelligibility benefit of one’s own recorded voice was significantly greater for participants who rated their own recorded voice higher in terms of similarity or intelligibility compared with those who rated it lower.

This paper aims to establish a method for preserving three-dimensional shape data of valuable wind instruments such as those housed in museums as cultural assets, as well as for creating replicas having the same playing feel and timbre as original instruments. In this paper, a three-dimensional computer model of a classical oboe made around 1800 was built through X-ray CT scanning, and physical copies of the instrument (replicas) were fabricated using an additive manufacturing. The replicas were then evaluated by analyzing the harmonic structure of the played sound and by listening experiments to see whether the timbres of the replicas could be distinguished from that of the original instrument. Despite the fact that the replicas were made of different materials from the original instrument, it was found that they had only slightly different harmonic structures from the original’s and that the subjects in the listening experiments had difficulty in distinguishing the timbres.

Thermoacoustic refrigerators offer the advantages such as being environmentally friendly due to the usage of common gases such as helium, argon, and nitrogen gases as their working fluids. This paper presents a nonlinear model of a traveling-wave thermoacoustic refrigerator capable of accurate analysis at large amplitudes, which enables the performance prediction and design of high-power thermoacoustic refrigerators. To develop the model, the loudspeaker parameters were measured and modeled. In addition, the acoustic power losses that occurred when sound waves passed through the bend and T-junction tubes were measured. Based on these experimental results, the thermoacoustic refrigerator model was proposed and verified under large-amplitude conditions.

This paper overviews neural target sound information extraction (TSIE), which consists of extracting the desired information about a sound source in an observed sound mixture given clues about the target source. TSIE is a general framework, which covers various applications, such as target speech/sound extraction (TSE), personalized voice activity detection (PVAD), target speaker automatic speech recognition (TS-ASR), etc. We formalize the ideas of TSIE and show how it can be implemented through various examples such as TSE, PVAD, and TS-ASR. We conclude the paper with a discussion of potential future research directions.

Ultrasound Computed Tomography is increasingly valued for breast cancer diagnosis. However, traditional echo imaging method, particularly the synthetic aperture method, though high in resolution, requires extensive data acquisition from numerous transducer element combinations. This leads to prolonged data collection, substantial storage and transfer burdens, and ultimately reduced throughput, limiting ultrasound computed tomography’s practicality. This study introduces a minimal data-acquisition strategy for echo imaging, leveraging a pixel-based imaging method. Our method aims to find an optimal transducer elements configuration that meets two criteria: minimizing the number of elements and ensuring the associated measurement matrix has a sufficiently small condition number to remain robust against random noise. By using this approach, we can significantly reduce the amount of data required for image reconstruction without compromising image quality compared to the synthetic aperture method. Through simulation studies, we demonstrate that our method requires only 1/8th the number of elements and 1/64th the amount of data used by the synthetic aperture method while reconstructing images with comparable resolution and enhanced contrast. This advancement is valuable in boosting screening throughput, lowering system costs, and potentially improving diagnostic accuracy, marking a significant step forward in ultrasound computed tomography technology for breast cancer detection.

The analysis of articulatory movements, particularly tongue movements, is a key challenge in speech research. However, traditional methods for extracting tongue contours often face difficulties due to poor image quality and noise interference, complicating tongue motion analysis. This study proposes a novel approach to automatically extract the contours of the tongue and other articulatory organs from ultrasound and real-time magnetic resonance imaging (rtMRI) data. We employed DeepLabCut (DLC), a deep-learning-based tool. Our experiments demonstrated that DLC is not reliant on image edges or contrast, demonstrating robustness against noise and enabling effective automatic contour extraction. This paper highlights the method used and evaluates the accuracy of contour extraction for the tongue and other articulatory organs. By leveraging advanced deep-learning techniques, we aim to enhance the understanding of articulatory movements and improve speech analysis tools, ultimately contributing to enhanced outcomes in speech therapy and pronunciation training.

In a previous paper, it was shown that the Kato model, with the material density and fiber diameter, can be used as a predictive model that enables the acoustic design (optimization of the fiber mixing rate) of fibrous materials. However, it does not have sufficient analytical precision for a material with high-bulk-density, so the model has limited applicability. Also, since the material density and fiber diameter are required as parameters, the Kato model cannot be applied to fibrous materials or foamed materials, the parameters of which are unknown. To solve these problems, in this paper, a technique of applying the Kato model to fibrous materials with high-bulk-density, fibrous materials with unknown material density and fiber diameter, and foamed materials is shown to extend its application.

We surveyed to explore how Japanese cochlear implant users perceive and enjoy music, as well as factors related to their music enjoyment. The survey gathered responses from 102 participants, showing that many continue to enjoy music after getting the implant and are keen on enjoying it even more. A time series analysis revealed that enjoyment of music decreased with hearing level, but after implantation, it improved to the same level as before the hearing loss. Additionally, there was a tendency for lower ratings of sound quality and music listening with increasing age, but participants with musical experience perceived music better than those without it. The study also suggested that practicing listening to music with cochlear implants may improve music perception. However, there were significant individual differences in the results. Especially after implantation, some people enjoyed music, and others did not. The findings indicate that future research should focus on enhancing subjective enjoyment of music.

Road traffic noise is one of the most widespread environmental noise sources, significantly impacting public health. To control traffic noise pollution, the European Union requires countries to prepare strategic road traffic noise maps using prediction models every five years since 2007. Similarly, Japan has developed its own road traffic noise maps based on regular on-site observations. However, traditional methods of collecting traffic data through on-site measurements are labor-intensive and costly. Therefore, in this work, we present a method for creating road traffic noise maps using an object detection deep learning algorithm to extract road traffic conditions, such as vehicle types and speeds, from aerial photographs. On the basis of ASJ RTN-Model 2018, the traffic condition data serves as the foundation for calculating the sound power levels of road traffic noise for different roads. The road traffic noise map is then created directly from the sound pressure level distribution within the considered regions. We validated the accuracy of the detection model and the calculated sound pressure level along the road from the aerial photographs against the published measurements. Using the proposed method, we created the road traffic noise map for Meguro City, Tokyo.

Ball and tire impact sources are used as standard heavy-weight impact sources to induce heavy-weight impact sound. To predict the impact sound insulation performance accurately, the modeling of impact sources is important. In this paper, new single-degree-of-freedom (SDOF) models are proposed. Numerical calculations validate that using the proposed SDOF models can predict the impact force exposure levels accurately. Then, a numerical example is used to investigate changes in impact force exposure levels when impact sources acting on the dry-type double floor system.

This paper proposes a filter selection algorithm of virtual sensing for feedback active noise control system tracking noise variations. The proposed system maintains noise reduction performance by switching to an optimal noise control filter according to variations in noise characteristics. Noise variations are detected using bandpass filters with different frequency responses, and the optimal noise control filter is selected based on these detections. The auxiliary filter in virtual sensing is then re-estimated to track the variation in the noise control filter. Simulation results with actual impulse responses show that the proposed system can maintain a noise reduction of approximately 10 dB even when the noise varies.

The perception of speaker’s facing directions is reflected radiation characteristics, which have an important role particularly in the immersive media. In the current study, we investigated the identification of speaker’s facing directions through two experiments: one with (Exp. A) and one without (Exp. B) differences in loudness, targeting the horizontal plane at 0˚, 45˚, 90˚, 135˚, and 180˚. The results of Exp. A suggested that participants mainly relied on loudness to identify speaker’s facing directions, whereas Exp. B suggested that identification could be judged from some acoustic cues such as the spectral component of the sound instead of loudness.

Sound wave propagation from a moving sound source, such as the Shinkansen, is affected not only by the source movement but also by the airflow around the source. Previous theoretical studies showed that variations in sound wave propagation depend on the airflow distribution. However, few experimental studies have been conducted, because the experimental study of the sound propagation from a moving sound source requires the installation of microphones, which inevitably affect the airflow. Thus, the actual effect of airflow remains unclear. In this study, we experimentally investigate the effect of airflow around a moving sound source using a parallel phase-shifting interferometer, which is a noncontact optical sound measurement method. We conducted wind tunnel tests with a scaled train model, which includes a sound source to visualize sound wave affected by a boundary layer. Additionally, we conducted moving-model tests to investigate the sound propagation around the moving scaled train model. The results show that the sound wave propagation characteristics change, and several phenomena, such as wavefront distortion, sound trapped in the boundary layer, and shadow zone formation, occur. Additionally, sound wavelength modulation and sound frequency modulation were observed in the wind tunnel and moving-model tests, respectively.

We developed a method for determining the inclusion of disturbance sounds in aircraft noise events using a convolutional neural network (CNN). Considering road-traffic noise as a disturbance sound, recognition models were developed for aircraft and road-traffic noise (hereafter referred to as the base model) and then combined. In addition, we developed a new model, the “frequency-split parallel model,” by advancing the base model. This approach involved inputting spectrograms split along the frequency axis. After verifying the accuracy of each model for single sound events of aircraft and road-traffic noises, the frequency-split parallel model was evaluated using superimposed data obtained around Narita International Airport and compared with the base model. The misidentification rate of road-traffic noise as aircraft noise decreased by approximately 57% compared to the base model, and 16 out of 17 measured superimposed events were correctly determined. These results confirm the effectiveness of the proposed method.

In ASJ RTN-Model 2018, a practical calculation method of the ground effect in A-weighted overall level for road traffic noise is specified by relatively simple formulae. The method is derived from the solution of wave propagation on an infinite and flat ground with acoustic impedance based on wave theory. By the practical calculation method, the attenuation level by the ground effect for a relatively long distance between the source and receivers tends to become larger on softer ground. Especially in the cases of lower sound source positions as specified in the ASJ RTN-Model 2018, the excess attenuation becomes considerably large. To investigate the appropriate treatment of the ground condition in the case of road traffic noise propagation on soft ground, the authors have performed an on-site measurement of road traffic noise. The measurement results were compared with calculation results by both the ASJ RTN-Model 2018 and the theoretical solution based on the wave theory. From the comparison, the features of the propagation characteristics obtained by the ASJ RTN-Model 2018 were discussed.

Sound fields around a barrier edge with an open-tube array are numerically calculated to investigate how the tube array affects barrier performance. Results show that the open tubes arrayed on the source side of the barrier edge are effective in some cases. On basis of the results, a hypothesis regarding the effects of the open tubes at the barrier edge is considered and tested. It is revealed that sound pressure levels behind the barrier are significantly suppressed when the phase difference between incident waves from the source to the upper and lower ends of a tube is small.

To evaluate the exterior noise reduction effects of electric vehicles (EVs) and hybrid electric vehicles (HEVs) under non-steady speed conditions, i.e., accelerating or decelerating running conditions, measurements of A-weighted sound power levels were conducted. This paper reports the results of a reanalysis using the data measured at the same test site. The results are shown as regression models of the sound power level depending on the vehicle speed. For internal combustion engine vehicles (ICEVs), there was a strong tendency for L_WA to increase when the acceleration was greater than 1.0 m/s². For EVs, when the acceleration is less than 2.0 m/s², which is considered the typical acceleration in general urban driving, it can be considered equivalent to the L_WA under steady-speed condition. During deceleration, there was no significant difference from steady-speed condition for either vehicle type. These differences were examined in terms of single-event sound exposure levels. In a case study that included stopping and going at a traffic signal, the difference in L_EA was shown to be 3.0 dB.

Both amplitude modulation and frequency modulation occur in a sound field radiated from moving sound sources. However, the amplitude modulation for different source types has not been clarified experimentally. In this study, the directivities of sound radiated from an ultrasonic transducer attached to a moving scaled train model and used as a simplified vibrational sound source were measured. The surface of the vibrational source was parallel to the movement direction of the source. The directivities of an aeolian tone sound generated using a cylinder attached to the train model were also measured; the cylinder was used as an aerodynamic sound source for a simplified dipole component of aerodynamic sound, which is dominant at low Mach numbers, generated from pressure fluctuation. The experimental results show that the acoustic intensity of sound radiated from a vibrational source increases roughly in proportion to the fourth power of the Doppler factor, whereas that of the dipole component of aerodynamic sound increases roughly in proportion to the eighth power of the Doppler factor. Sound radiated from an aerodynamic source moving at low Mach numbers leads to a greater amplitude modulation due to source movement compared with sound radiated from a vibrational source.

There have been only a limited number of studies in which the subjective responses to traffic-induced vibration and noise in buildings are experimentally compared among participants from different nations. In this study, we investigate the subjective responses of two groups to vibration and noise: one group comprises 20 Vietnamese participants and the other 20 Japanese participants. The participants were exposed to noises of six different levels and vibrations of five varying magnitudes, which were obtained from field measurement recordings of road traffic, Shinkansen railways, and conventional railways. They were asked to rate their levels of discomfort and reading disturbance on a seven–ordinal scale and a five–ordinal scale, respectively. The results indicate that the Japanese participants rated significantly higher discomfort levels in response to noises of most levels and vibrations of certain magnitudes, than the Vietnamese participants. The difference in reading disturbance level between the two groups was less clear than that in discomfort level. The subjective equivalences between vibration and noise did not differ significantly between the Japanese and Vietnamese groups.

Areas in the vicinity of trunk roads are exposed to high levels of noise and may pose high health risks to residents. To assess the health risks and formulate effective noise mitigation measures, prediction of road traffic noise is crucial. Addressing this issue, the author has developed a road traffic noise prediction system that allows prediction of sound levels using a database of road network. The objective of the present study was to predict road traffic noise in the vicinity of the trunk roads using a Digital Road Map Platform (DRM-PF) database, which contains nationwide road geometries and traffic settings of trunk roads. Predicted sound levels were compared with actual measurements and noise maps were created to demonstrate the feasibility of assessing noise exposure and associated health risks in the vicinity of trunk roads in Japan. The results show a generally good agreement between predicted and measured levels, while challenges remain in accurate prediction in a number of environments, mainly due to the lack of accurate geometries. The extensive coverage of the DRM-PF database throughout Japan enables noise mapping in arbitrary regions near trunk roads, which would contribute to making noise policy.