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

作者：Abe Tomomi^*; Hashimoto Shuji; Matsumoto Mitsuharu

来源：Journal of the Acoustical Society of America, 2010, 127(2): 896-901.

DOI：10.1121/1.3278601

• acoustic correlation
• filtering theory
• signal denoising
• speech processing

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.

• 出版日期2010-2