With rising demand for sensor networks in recent years, microelectromechanical systems (MEMS) sensors are growing more attractive because they are tiny and available at increasingly lower cost. A data sample detected using a MEMS sensor is, in general, stored temporarily in a register of the sensor module. It is regarded as a buffer of only one sample. Therefore, buffer overrun/underrun is likely to occur. A precise method of estimating update interval of the register is discussed herein in conjunction with modeling of the problem. According to the comparison between every pair of adjacent data samples at various intervals, binary random variables are constructed. Based on the proposed model, the update interval is estimated by a slightly modified version of basic estimation algorithms such as ML and EM. It is shown that the proposed method successfully estimates the update intervals for an actual MEMS pressure sensor.

A method is proposed for estimating the source signal and its direction of arrival (DOA) in this paper. It is based on ML estimation of the transfer function between microphones combined with the EM algorithm for a Gaussian Mixture Model (GMM), assuming that the signal is captured at each microphone with delay corresponding to the traveling of sound and some decay. By this modeling, search for the maximum log-likelihood in the ML estimation can be realized simply by eigenvalue decomposition of a properly designed matrix. Computer simulation results show that the proposed method achieves SDR of greater than 10 dB regardless of amplitude difference between microphones and DOA estimation error of less than 8 degrees, on average. It is also shown that it can maintain high performance in various conditions.

Conversion of microphone array signals into loudspeaker array signals is an essential process in high-definition spatial audio. This paper presents the theory of boundary matching filters (BMFs) for spherical array signal conversion. BMFs adapt the physical boundary conditions used during recording to the ones required for reproduction by relying on a theoretical framework provided by the Kirchhoff-Helmholtz integral equation (KHIE). Computationally, array signal conversion is performed in a transform domain where sound fields are represented in terms of spherical harmonic functions. Related research on transform-domain signal conversion filters is interpreted in the context of the KHIE. The case of a rigid recording boundary and an open reproduction boundary is addressed. The proposed rigid-to-open BMFs provide a suitable basis for designing gain-limited filters to deal with the problem of excessive gains at certain frequency bands, observed when using high-resolution arrays. Spatial, spectral, and temporal effects in sound field reconstruction when finite numbers of transducers are used in anechoic conditions are investigated analytically and exemplified numerically. Results show that the proposed gain-limited rigid-to-open BMFs outperform the existing gain-limited filters based on Tikhonov regularization because they reduce the spatial discretization effects and yield impulse responses that are more localized around their main peaks.

For sound field reproduction that includes height (with-height reproduction), it is more efficient to record and reproduce the sound field with lower resolution in elevation than in azimuth due to the spatial abilities of human auditory perception. We propose a sound field reproduction method using horizontally arranged cylindrical arrays of microphones and loudspeakers, which is based on the wave field reconstruction (WFR) filter. With the use of cylindrical array configurations, it is possible to reproduce sound waves arriving from upper and lower directions with a smaller number of array elements at angular positions. The WFR filter is analytically derived in the cylindrical harmonic domain and allows direct transformation from the received signals of the microphones into the driving signals of the loudspeakers. A model in which microphones are mounted on a rigid cylindrical baffle is introduced to stabilize the WFR filter. Numerical simulation results indicated that the reproduction accuracy in the neighboring region along the central axis of the cylindrical array was better preserved when using the proposed method than when the method with planar arrays was used.

For with-height reproduction of a sound field, an efficient technique is to record and reproduce the sound field at lower resolution at an elevation angle than that at a horizontal angle based on auditory perception. To achieve this, developing a method using multiple horizontal linear arrays of microphones and loudspeakers is necessary. We propose a sound field reproduction method for a cylindrical array configuration which is based on a wave field reconstruction (WFR) filter analytically derived in the helical wave spectrum domain. The filter is stabilized by introducing a model in which microphones are mounted on a rigid cylindrical baffle. Numerical simulation results indicated that the reproduction accuracy of the proposed method was well preserved in the near-field region along the central axis of the cylinder even when the number of elements at angular position is small.

We can perceive a sound position from binaural signals using mainly head-related transfer functions (HRTFs). Using the theorem presented herein, we can display a sound image to a specific position in virtual auditory space by HRTFs. However, HRTF is defined commonly in a free-field, and a virtual sound image is perceived as a dry source without reflection, reverberation, or ambient noise. Therefore, the virtual sound space might be unnatural. The authors developed a software-based virtual auditory display (VAD) that outputs audio signals for a set of headphones with a three-dimensional position sensor. The VAD software can display a dynamic virtual auditory space that is responsive to a listener's head movement. Subjective evaluations were conducted to clarify the relation between the perceived reality of virtual sound space and ambient sound. Evaluation results of the reality of the virtual sound space displayed by the VAD software are introduced.

In this paper, we propose a novel data hiding technique for G.711 speech based on the LSB substitution method. The novel feature of the proposed method is that a low-bitrate encoder, G.726 ADPCM, is used as a reference for deciding how many bits can be embedded in a sample. Experiments showed that the method outperformed the simple LSB substitution method and the selective embedding method proposed by Aoki. We achieved 4-kbit/s embedding with almost no subjective degradation of speech quality, and 10 kbit/s while keeping good quality.

Conventional watermarking techniques based on echo hiding provide many benefits, but also have several disadvantages, for example, a lenient decoding process, weakness against multiple encoding attacks, etc. In this paper, to improve the weak points of conventional echo hiding, we propose a time-spread echo as an alternative to the single echo in conventional echo hiding. Spreading an echo in the time domain is achieved by using pseudonoise (PN) sequences. By spreading the echo, the amplitude of each echo can be reduced, i.e., the energy of each echo becomes small, so that the distortion induced by watermarking is imperceptible to humans while the decoding performance of the embedded watermarks is better maintained as compared with the case of conventional echo hiding, as shown by computer simulations, in which several parameters, such as the amplitude and length of PN sequences and analysis window length, were varied. Robustness against typical signal processing was also evaluated in these simulations and showed fair performance. Results of a listening test using some pieces of music showed good imperceptibility.

A pitch-scaling modification of a watermarked signal is a severe attack on audio watermarking techniques. In particular, watermarking techniques using a pseudonoise (PN) sequence as a secret key are fragile in the face of this modification because of the desynchronization between the original and embedded PN sequences. We, therefore, propose a log-scaling watermark detection method to cope with this problem. This method is based on a general logarithm operation in the frequency or quefrency axis in the decoding process. This method is also advantageous in that no other modification of the original embedding process is necessary. The validity of the proposed log-scaling watermark detection method is evaluated by employing the time-spread echo method as an example of such a watermarking technique.

Isolated-monosyllable listening tests are widely used in Japan to assess personal hearing ability. However, word intelligibility tests could be more suitable for measuring hearing ability. We developed new word lists with controlled word familiarity and phonetic balance. Listening test results showed slight differences in intelligibility scores among word lists within the same word familiarity rank. We examined whether such variation can be compensated by changing the signal to noise ratio (SNR) of each word to equalize intelligibility scores of word lists in identical word familiarity ranks. Results of listening tests indicated that a 1.0 difference in word familiarity was approximately equivalent to a 2.0 dB difference in SNR. Based on the results, we developed and tested a method of compensating for the difference caused by word familiarity by controlling SNR. We confirmed that this compensation method was effective in equalizing the intelligibility score of each word list.

Conventional watermarking techniques based on echo hiding provide many benefits, but with several weak points, for example, a lenient decoding process, weakness against multiple encoding attack, etc. In this paper, to improve the weak points of conventional echo hiding, we propose a time-spread echo as an alternative to the conventional echo in the echo hiding. Spreading an echo is performed in the impulse response domain using PN sequences. We also evaluate the characteristics of the proposed method as a function of two parameters: gain and length of PN sequences used in spreading an echo. The proposed method is expected to provide more natural and colorless sounds as well as better imperceptibility of distortion as compared with conventional echo hiding.

Two major objectives of a binaural microphone array system are reducing background noise and retaining interaural relationships, for example, ILD and ITD. Hence, a new adaptive microphone array signal processing based on a weighted least squares algorithm for binaural hearing is proposed in this paper. Instead of Lagrange's method of indetenninant coefficients, which is often used in conventional binaural microphone array systems, a weighted least squares algorithm is employed. Results of computer simulations carried out for purposes of evaluation showed that the proposed method is characterized by flexible and robust adaptivity to various conditions.

In active control systems, the configuration of the control sources greatly affects the performance. In this paper, a method for optimizing the number and location of control sources is proposed. As a criterion for the optimization, the linear independence of the transfer impedance vector is employed. A physical interpretation of the linear independence of the transfer impedance vector is the independent contribution of each control source that is not substitutable by the other sources. In the optimization process, control sources are determined one by one using Gram-Schmidt orthogonalization so that the corresponding transfer impedance vectors are the most linearly independent. The results of the simulation show the optimized configuration to have small control error. Moreover, for the optimized configuration, the transfer impedance matrix exhibits a small condition number, which results in robustness of the system in the face of changes in the environment.

A method for optimizing the number and the configuration of control sources in an active control system is proposed. In the optimization process, sources are selected one by one so that the corresponding transfer impedance vector is the most linearly independent. From the results of the simulation, it is shown that the optimized configuration yields not only a small average control error but also a small condition number in the transfer impedance matrix, which contributes to the robustness of the system against environmental change.

For the effective reduction of noise that is suddenly switched on or transient, attenuation is required from its onset. A weighted recursive least square approach, where an extra term providing an initial directivity is combined with the power minimization term, is proposed. A noise reduction system is also proposed.>