Noise stripping of speech signal is an important topic for maintaining the stable speech communication circuit. SPAC(Speech Processing system by use of Auto Correlation function)skillfully utilizing the nature of correlation function is considered effective for this purpose. This paper treats at the signal to noise ratio from -10 to 10dB. A simulation of SPAC reveals that the noise level reduction and improvement in the signal to noise ratio are successfully made. The quality of speech processed by SPAC is evaluated by hearing tests such as a pair comparison test, an opinion test of five categories and a syllable articulation test. It is shown that SPAC is subjectively effective for noise level reduction though no improvement in the articulation score is achieved.
As is well-known, the internal physical mechanism of noise generation and propagation which directly have a great influence on the noise fluctuation form, such as the distributions of location of vehicles, the distance between two arbitrary vehicles flowing in the same direction and the number of vehicles in a certain segment on road and the sound propagation environment are most fundamental and important in order to investigate the statistical properties of road traffic noise. Up to now, many approaches for estimating the noise intensity distribution function have been carried out in an idealized case with an exponentially-distributed vehicles model. However, an actual road traffic flow shows too complex pattern to be exactly explained by the above simplified model. In this paper, firstly, we have focused discussions on the prediction problem of distribution form of road traffic noise fluctuation in the near sound field which is important from the viewpoint of noise environmental pollution problem, and we theoretically have derived an explicit expression of probability distribution function and several types of its simplified expression form applicable to the arbitrary road traffic flow containing the idealized traffic model as a special case. Next, we have discussed the relations between theoretical results and well-known previous studies, and shown that our theory includes the well-known previous results as special cases. Finally, we have applied our theory to actual road traffic noise observed in Hiroshima city for purpose of confirming experimentally the legitimacy of our theory.