Ryoichi MIYAZAKI Hiroshi SARUWATARI Kiyohiro SHIKANO
We propose a structure-generalized blind spatial subtraction array (BSSA), and the theoretical analysis of the amounts of musical noise and speech distortion. The structure of BSSA should be selected according to the application, i.e., a channelwise BSSA is recommended for listening but a conventional BSSA is suitable for speech recognition.
This letter presents a new automatic musical genre classification method based on an informative song-level representation, in which the mutual information between the feature and the genre label is maximized. By efficiently combining distance-based indexing with informative features, the proposed method represents a song as one vector instead of complex statistical models. Experiments on an audio genre DB show that the proposed method can achieve the classification accuracy comparable or superior to the state-of-the-art results.
Kenji TAGUCHI Tatsuya KASHIWA Kohzoh OHSHIMA Takeshi KAWAMURA
Inter-vehicle communication (IVC) system using 700 MHz band to prevent car crashes has been proposed recently. In this paper, we first apply the FDTD method to the analyses of propagation characteristics at an intersection for IVC. We investigate the propagation characteristics considering the electrical conductivities, thickness and windows of building wall and pedestrians. As a result, it is shown that the electrical conductivities and thickness of building wall have a slight influence. In contrast, windows and pedestrians have a great influence on the propagation characteristics. Furthermore, the azimuth delay profiles are obtained by using the MUSIC algorithm.
In this study, a discriminative weight training is applied to a support vector machine (SVM) based speech/music classification for a 3GPP2 selectable mode vocoder (SMV). In the proposed approach, the speech/music decision rule is derived by the SVM by incorporating optimally weighted features derived from the SMV based on a minimum classification error (MCE) method. This method differs from that of the previous work in that different weights are assigned to each feature of the SMV a novel process. According to the experimental results, the proposed approach is effective for speech/music classification using the SVM.
Takashi MIWA Shun OGIWARA Yoshiki YAMAKOSHI
Recently, it has become important to rapidly detect human subjects buried under collapsed houses, rubble and soil due to earthquakes and avalanches to reduce the casualties in a disaster. Such detection systems have already been developed as one kind of microwave displacement sensors that are based on phase difference generated by the motion of the subject's breast. Because almost all the systems consist of single transmitter and receiver pair, it is difficult to rapidly scan a wide area. In this paper, we propose a single-frequency multistatic radar system to detect breathing human subjects which exist in the area surrounded by the transmitting and receiving array. The vibrating targets can be localized by the MUSIC algorithm with the complex amplitude in the Doppler frequency. This algorithm is validated by the simulated signals synthesized with a rigorous solution of a dielectric spherical target model. We show experimental 3D localization results using a developed multistatic Doppler radar system around 250 MHz.
Yung-Yi WANG Wen-Hsien FANG Jiunn-Tsair CHEN
We propose a dimension reduction algorithm for the receiver of the downlink of direct-sequence code-division multiple access (DS-CDMA) systems in which both the transmitters and the receivers employ antenna arrays of multiple elements. To estimate the high order channel parameters, we develop a layered architecture using dimension-reduced parameter estimation algorithms to estimate the frequency-selective multipath channels. In the proposed architecture, to exploit the space-time geometric characteristics of multipath channels, spatial beamformers and constrained (or unconstrained) temporal filters are adopted for clustered-multipath grouping and path isolation. In conjunction with the multiple access interference (MAI) suppression techniques, the proposed architecture jointly estimates the direction of arrivals, propagation delays, and fading amplitudes of the downlink fading multipaths. With the outputs of the proposed architecture, the signals of interest can then be naturally detected by using path-wise maximum ratio combining. Compared to the traditional techniques, such as the Joint-Angle-and-Delay-Estimation (JADE) algorithm for DOA-delay joint estimation and the space-time minimum mean square error (ST-MMSE) algorithm for signal detection, computer simulations show that the proposed algorithm substantially mitigate the computational complexity at the expense of only slight performance degradation.
In this letter, we propose a novel approach to speech/music classification based on the support vector machine (SVM) to improve the performance of the 3GPP2 selectable mode vocoder (SMV) codec. We first analyze the features and the classification method used in real time speech/music classification algorithm in SMV, and then apply the SVM for enhanced speech/music classification. For evaluation of performance, we compare the proposed algorithm and the traditional algorithm of the SMV. The performance of the proposed system is evaluated under the various environments and shows better performance compared to the original method in the SMV.
Koichi ICHIGE Kazuhiko SAITO Hiroyuki ARAI
This paper presents a high resolution Direction-Of-Arrival (DOA) estimation method using unwrapped phase information of MUSIC-based noise subspace. Superresolution DOA estimation methods such as MUSIC, Root-MUSIC and ESPRIT methods are paid great attention because of their brilliant properties in estimating DOAs of incident signals. Those methods achieve high accuracy in estimating DOAs in a good propagation environment, but would fail to estimate DOAs in severe environments like low Signal-to-Noise Ratio (SNR), small number of snapshots, or when incident waves are coming from close angles. In MUSIC method, its spectrum is calculated based on the absolute value of the inner product between array response and noise eigenvectors, means that MUSIC employs only the amplitude characteristics and does not use any phase characteristics. Recalling that phase characteristics plays an important role in signal and image processing, we expect that DOA estimation accuracy could be further improved using phase information in addition to MUSIC spectrum. This paper develops a procedure to obtain an accurate spectrum for DOA estimation using unwrapped and differentiated phase information of MUSIC-based noise subspace. Performance of the proposed method is evaluated through computer simulation in comparison with some conventional estimation methods.
Qiaowei YUAN Qiang CHEN Kunio SAWAYA
MUSIC-based estimation of direction of arrival (DOA) using universal steering vector (USV) is experimentally studied. A four-element array antenna and a four-channel receiver are employed for the experiment. In order to improve the accuracy of DOA estimation, USV which has already included the effect of mutual coupling between array elements and effect of array elements themselves is compensated to further include the electric delay and loss of four channels in the receiver. The compensated USV (C-USV) approach proposed in this paper does not need the time-consuming measurement of array element pattern because the compensating matrix for USV is obtained by measuring the S parameters between RF input ports of the feeding cables and IF output ports of the receiver. The experimental results of MUSIC-based DOA estimation show that C-USV approach is an accurate, effective and practical method for the MUSIC-based DOA estimation.
Koichi ICHIGE Yoshihisa ISHIKAWA Hiroyuki ARAI
This paper presents a simple but high resolution DOA estimation method using second-order differential of MUSIC spectrum. MUSIC method is paid attention as one of "Superresolution" DOA estimation methods because of their brilliant characteristics, however MUSIC also has the problem of estimation accuracy in severe environments like low SNR, small number of snapshots, or incident waves from closely-spaced angles. Especially the case of two or more incident waves from closely-spaced angles, MUSIC often fails in making spectrum peaks that leads inaccurate DOA estimation. We pay attention to the fact that the second-order differential of MUSIC spectrum makes negative peaks around the original DOAs even when MUSIC spectrum does not make peaks there. We try to estimate DOAs not by MUSIC spectrum but by the second-order differential of the MUSIC spectrum, and to find its peaks for being estimated DOAs. The performance of the present method is evaluated in compared with MUSIC and Root-MUSIC methods through computer simulations and experiments.
This paper presents a low cost and portable DOA (Direction Of Arrival) estimation system for surveillance using a modifed beamspace MUSIC (MUltiple Signal Classification) by a quasi-orthogonal multi-beam. This is instead of DFT processing and hardware system consisting of chip-sized phase shifters, a single ADC (Analogue to Digital Converter) and a single TR (TRanceiver) module for an antenna array. In the beamspace MUSIC, generated beampatterns have orthogonal properties. This proposed system cannot make such a beampattern due to the variable range limitation of phase shifter, then we use the quasi-orthogonal beam obtained by the calculation of correlation coefficient for beampattern. We demonstrate the proposed system using 4-element microstrip array antenna and chip-sized phase shifters. The DOA experiment in anechoic chamber confirms the proposed system performance.
A one dimensional (1-D) based tree structure algorithm is proposed for estimating the 2D-DOAs of the signals impinging on a uniform rectangular array. The key idea of the proposed algorithm is to successively utilize the 1-D MUSIC algorithm several times, in tree structure, to estimate the azimuth and the elevation angles independently. Subspace projectors are exploited in conjunction with the 1-D MUSIC algorithms to decompose the received signal into several signals each coordinated by its own 2D-DOA. The pairing of the azimuth estimates and the associated elevation estimates is naturally determined due to the tree structure of the algorithm.
Yoshihisa ISHIKAWA Koichi ICHIGE Hiroyuki ARAI
This paper presents a scheme for accurately detecting the number of incident waves arriving at array antennas. The array antenna and MIMO techniques are developing as 4th generation mobile communication systems and wireless LAN technologies, and the accurate estimation of the propagation environment is becoming more important. This paper emphasizes the accurate detection of the number of incident waves; one of the important characteristics in multidirectional communication. There are some recent papers on accurate detection but they have problems of estimation accuracy or computational cost in severe environment like low SNR, small number of snapshots or waves with close angles. Hence, AIC and MDL methods based on statistics and information theory are still often used. In this paper, we propose an accurate estimation method of the number of arrival signals using the orthogonality of subspaces derived from preliminary estimation of signal subspace. The proposed method accurately estimates the number of signals also in severe environments where AIC and MDL methods can hardly estimate. We evaluate the performance of these methods through some computer simulation and experiments in anechoic chamber.
Mansoo PARK Hoi-Rin KIM Yong Man RO Munchurl KIM
The noise robustness of an audio fingerprinting system is one of the most important issues in music information retrieval by the content-based audio identification technique. In a real environment, sound recordings are commonly distorted by channel and background noise. Recently, Philips published a robust and efficient audio fingerprinting system for audio identification. To extract a robust and efficient audio fingerprint, Philips applied the first derivative (differential) to the frequency-time sequence of the perceptual filter-bank energies. In practice, however, the noise robustness of Philips' audio fingerprinting scheme is still insufficient. In this paper, we introduce an extension method of the audio fingerprinting scheme for the enhancement of noise robustness. As an alternative to frequency filtering, a type of band-pass filter, instead of a high-pass filter, is used to achieve robustness to background noise in a real situation. Our experimental results show that the proposed filter improves the noise robustness in audio identification.
Takuya YOSHIOKA Takafumi HIKICHI Masato MIYOSHI Hiroshi G. OKUNO
This paper describes a method for estimating the amplitude characteristics of poles common to multiple room transfer functions from musical audio signals received by multiple microphones. Knowledge of these pole characteristics would make it easier to manipulate audio equalizers, since they correspond to the room resonance. It has been proven that an estimate of the poles can be calculated precisely when a source signal is white. However, if a source signal is colored as in the case of a musical audio signal, the estimate is degraded by the frequency characteristics originally contained in the source signal. In this paper, we consider that an amplitude spectrum of a musical audio signal consists of its envelope and fine structure. We assume that musical pieces can be classified into several categories according to their average amplitude spectral envelopes. Based on this assumption, the amplitude spectral envelope of the musical audio signal can be obtained from prior knowledge of the average amplitude spectral envelope of a musical piece category into which the target piece is classified. On the other hand, the fine structure is identified based on its time variance. By removing both the spectral envelope and the fine structure from the amplitude spectrum estimated with the conventional method, the amplitude characteristics of the acoustical poles can be extracted. Simulation results for 20 popular songs revealed that our method was capable of estimating the amplitude characteristics of the acoustical poles with a spectral distortion of 3.11 dB. In particular, most of the spectral peaks, corresponding to the room resonance modes, were successfully detected.
Takashi KATO Kazumasa TAIRA Kunio SAWAYA Risaburo SATO
An estimation method of source location of undesired electromagnetic wave from electronic devices by using the MUSIC algorithm is proposed. The MUSIC algorithm can estimate the direction of arrival accurately, however, the estimation error is large in the case of short range multiple coherent sources. In order to overcome this problem, a method to improve the estimation accuracy is presented. Experimental results show that the proposed method can reduce the maximum estimation error from 7 cm of the conventional method to 2 cm.
Audrey BLIN Shoko ARAKI Shoji MAKINO
This paper focuses on the underdetermined blind source separation (BSS) of three speech signals mixed in a real environment from measurements provided by two sensors. To date, solutions to the underdetermined BSS problem have mainly been based on the assumption that the speech signals are sufficiently sparse. They involve designing binary masks that extract signals at time-frequency points where only one signal was assumed to exist. The major issue encountered in previous work relates to the occurrence of distortion, which affects a separated signal with loud musical noise. To overcome this problem, we propose combining sparseness with the use of an estimated mixing matrix. First, we use a geometrical approach to detect when only one source is active and to perform a preliminary separation with a time-frequency mask. This information is then used to estimate the mixing matrix, which allows us to improve our separation. Experimental results show that this combination of time-frequency mask and mixing matrix estimation provides separated signals of better quality (less distortion, less musical noise) than those extracted without using the estimated mixing matrix in reverberant conditions where the reverberant time (TR) was 130 ms and 200 ms. Furthermore, informal listening tests clearly show that musical noise is deeply lowered by the proposed method comparatively to the classical approaches.
Toshihiko FUKUE Atsushi FUJITA Nozomu HAMADA
In this paper we propose a stepped-FM array radar system that can precisely estimate the target position by combining S- and T-MUSIC and adaptive beamforming. By adopting the adaptive beamformer as a preprocessor of T-MUSIC, the proposed system can uniquely determine the direction and distance of targets. In addition, the distance estimation precision is improved by introducing beamformer.
Masaki TAKANASHI Toshihiko NISHIMURA Yasutaka OGAWA Takeo OHGANE
Mainly, a uniform linear array (ULA) has been used for DOA estimation of coherent signals because we can apply the spatial smoothing preprocessing (SSP) technique. However, estimation by a ULA has ambiguity due to the symmetry, and the estimation accuracy depends on the DOA. Although these problems can be solved by using a uniform circular array (UCA), we cannot estimate the DOA of coherent signals because the SSP technique cannot be applied directly to the UCA. In this paper, we propose to estimate 2-dimensional DOA (polar angles and azimuth angles) estimation of coherent signals using a cylindrical array which is composed of stacked UCAs.
To understand radio propagation structures and consider signal recovering techniques in mobile communications, it is most effective to estimate the signal parameters (e.g., DOA) of individual incoming waves. Also, in radar systems, it is required to discriminate the desired signal from interference. As one of the high-resolution DOA estimators, MUSIC and ESPRIT have attracted considerable attention in recent years. They need the eigenvectors of the correlation matrix and therefore we have to execute the EVD (eigenvalue decomposition) of correlation matrix. However, the EVD generally brings us a heavy computational load and as a result it is difficult to realize the real-time DOA estimator, which will be useful as a multibeam-forming algorithm for adaptive antennas. This paper focuses on MUSIC and ESPRIT using subspace tracking methods, such as BiSVD, PAST, and PASTd, to carry out iterative DOA estimation. Then, they are compared through computer simulation. Adaptive beamforming based on DCMP and MLM is also mentioned and an example is shown.