A new method for recovering a broad band reverberant speech is proposed and evaluated, which uses a sub-band multi-channel inverse filter based on the multiple-input/output inverse theorem (MINT) proposed by M.Miyoshi and Y.Kaneda. The MINT method requires solving simultaneous linear equations of tens of thousands orders in the case of the reverberation time is 1 second and the sampling rate is 20 kHz. To overcome the numerical problem, we propose a sub-band MINT (SB-MINT method), in which the full band speech signal width (OwlOkHz) is divided into 512 sub-bands and each of the sub-band is processed by the MINT method to recover the effect of the reverberation. The SB-MINT is evaluated for , various reverberation conditions, and is found to be very effective for recovery of broad band reverberant speech.

No abstract

Owing to the rapid development of electronic and photonic devices operating in terahertz (THz) frequency region, it is crucial to understand the non-thermal effects of these electromagnetic waves on human body. In the past, it was reported that human sweat duct works as a helical antenna in terahertz frequency region and it plays an important role in the interaction of THz wave with the skin. In order to further understand the effects of these waves, it is essential to know the morphological features of the sweat ducts. In this work, we analyzed the density and dimension of the ducts based on three dimensional images obtained with optical coherence tomography (OCT). We observed that the sweat ducts possess helical structure and the diameter of the ducts from different regions of the body show considerable resemblance. Based on this information, we numerically determined the resonating frequency of helical ducts which resulted in the terahertz frequency region spanning from 0.3 THz to 0.7 THz.