Multichannel speech separation and localization by frequency assignment

The Journal of the Acoustical Society of America

2008

This paper describes an acoustical array combining microphones and piezoelectric devices. Conventional microphone arrays have been widely utilized to realize noise reduction, sound separation and direction of arrival estimation system. However, when a conventional microphone array is mounted on a real system, such as a machine, vehicle or robot, the microphones are set extremely close to the system's actual body. In such cases, the noise from the system itself, such as motors, gears, and engines, namely internal noise, often becomes a troublesome problem. It is difficult to reduce internal noise utilizing a conventional microphone array because internal noise sources are extremely close to the microphones. As internal noise is not always stationary, statistically independent or sparse, most useful blind source separation approaches, such as independent component analysis and the sparseness approach, cannot be employed. Our aim is to reduce internal noise utilizing microphones and piezoelectric devices attached to the internal noise source. In this paper, a general description of the acoustical array is formulated and the characteristic features of microphones and piezoelectric devices in an acoustical array are given. An acoustical array combining microphones and piezoelectric devices is also described with some experimental results.

Section: Introductionmentioning

confidence: 99%

An acoustical array combining microphones and piezoelectric devices

The Journal of the Acoustical Society of America

2008

“…When the number of the noise sources is larger than that of the microphones, ICA also cannot be employed. Some approaches utilize sparseness to solve the BSS problem [15], [16], [17]. Sparseness means that most of the frequency components of a signal are zero, that is, the sources rarely overlap in the frequency domain.…”

Section: Introductionmentioning

confidence: 99%

Noise reduction combining microphones and laser listening devices

Abe

2008 IEEE International Conference on Mechatronics and Automation

2008

This paper describes noise reduction combining microphones and laser listening devices. A microphone array is one of the useful approaches for reducing the noise. However, when the microphones are mounted on robot systems, the problem of internal noise from robots such as motors and gears arises. It is difficult to reduce the internal noise utilizing the conventional microphone array because the noise source is extremely close to the microphones. As the internal noise is not always stationary or sparse, some useful blind source separation (BSS) approaches such as Independent Component Analysis (ICA) and sparseness approach cannot be employed. Our aim is to reduce the internal noise by a sensor fusion of microphones and laser listening devices. In this paper, we describe the problem of the typical microphone array and show the design of the laser listening devices. Noise reduction combining microphones and laser listening devices is then described. Experimental results are also given to show the effectiveness of the proposed method.

“…There have been many studies about multichannel signal processing for noise reduction in acoustical signal processing such as microphone array, 1-3 independent component analysis, 4,5 and sparseness approaches. [6][7][8] However, when we consider downsizing the system, single-channel approaches have several advantages compared to multichannel approaches. An approach utilizing a single-channel signal can also be simpler than one employing a multichannel system.…”

Section: Introductionmentioning

confidence: 99%

Automatic parameter optimization in epsilon-filter for acoustical signal processing utilizing correlation coefficient

Abe

The Journal of the Acoustical Society of America

2010

epsilon-filter can reduce most kinds of noise from a single-channel noisy signal while preserving signals that vary drastically such as speech signals. It can reduce not only stationary noise but also nonstationary noise. However, it has some parameters whose values are set empirically. So far, there have been few studies to evaluate the appropriateness of the parameter settings for epsilon-filter. This paper employs the correlation coefficient of the filter output and the difference between the filter input and output as the evaluation function of the parameter setting. This paper also describes the algorithm to set the optimal parameter value of epsilon-filter automatically. To evaluate the adequateness of the obtained parameter, the mean absolute error is calculated. The experimental results show that the adequate parameter in epsilon-filter can be obtained automatically by using the proposed method.