FEEDBACK ANC AND SPEECH ENHANCEMENT FOR ACTIVE HEARING PROTECTION ON THE EXTREME NOISY INDUSTRY ENVIRONMENT

摘要

In this paper, we propose an active hearing protection system for workers in an extreme noise industrial environment. For the active hearing protection, we utilize ANC (Active and speech enhancement technology. Active hearing protection provides face to face communication that voice is passed through and the noise can be blocked. For this object we propose a method of estimating a secondary path of ANC accordance with user's hearing protection wearing characteristics and a speech enhancement algorithm to improve speech intelligibility with the ANC and the De-Noised speech signal. An adaptive RMC (Residual Music Canceler) is proposed for enhancing the accuracy of the reference signal of the feedback ANC. We obtained results that the secondary path can be accurately estimated and high quality of music sound can be consistently obtained in a time-varying secondary path situation. Most of the ANC system use the LMS algorithm to design a secondary path filter. However the convergence time of the filter is prolonged in the conventional method because the initial coefficient values of the secondary path filter is set to O.To solve this problem, in this paper we propose a method of shortening the convergence time of the filter by setting the coefficient values determined in advance as an initial value of the secondary pass filter. With the proposed method the error is decreased from the initial stage of the secondary path filter, whereby the convergence of the ANC system is increased. In the experiments by using a commercially available headset it was confirmed that the proposed method is faster than the conventional method for convergence of ANC system. By applying the ANC and the De-Noising system in real-time DSP (Digital Signal Processor), we have developed an active hearing protection.