| Mask effect is the phenomenon that when two or more sounds exist together, the weaker sound is made inaudible in the presence of a louder sound. Since 1930s of last century, the effect has been studied widely in abroad. In recent decades, because of its usage in the fields of audio signal compression,transmission and processing, including in MPEG和Dolby Digital (AC-3) audio compression standard, DVD digital audio broadcasting ,internet and IP telephone etc. The effect became an international hot topic and many new achievements have been reached. In the first, the dissertation studied the relationship between MDCT and simultaneous mask effect curve。It is pointed out that although the MDCT coefficient XMDCT (k)represents a sinusoidal tone, XMDCT (k) is different from DFT coefficient XDFT (k)that it cannot represent the amplitude of a sinusoidal component by itself alone and the corresponding analog frequency of the former is larger than that of the latter. Because the change of amplitude and the shift of analog frequency of the sinusoidal component represented by XMDCT (k) are sample ratio and transform block size related, it is impossible to get a MDCT simultaneous mask effect curve adapting to all sample ratio and transform block size by experiment or theory analysis. To calculate the simultaneous mask threshold when quantifying the MDCT coefficient, the DFT is still needed.Then,simultaneous mask effects when tones were masked with different SPL(Sound Pressure Level)narrow-band pink noises had been studied. The experiment results and analysis showed that when the SPL of the masker was 60dB, the average value of up slope was 21.6 dB/Bark, and the down slope was -20.7dB/Bark. With increasing of masker's SPL,the average value of up slope increased but the down slope almost kept unchanged。At the same time, the scope of mask effect expanded within critical band, which meant the mask peak became shallower, and on the high frequency side the slope of the curves became shallower too. Additionally,the slope of the curve did not increase or decrease obviously with the increasing of frequency. The main causes of above results were propagation mechanism of nerve fiber and two-tone suppression.The paper also studied the influence of the masker&maskee's spatial separation in azimuth on simultaneous mask effect. The results of theoretical analysis and experiment showed that when the masker and maskee were spatially separated, the thresholds tended to drop and produced the phenomena of spatial unmasking. The results also showed that the spatial unmasking was frequency related. The higher the frequency was, the stronger the spatial unmasking was. The main reason was that when a masker and a maskee source were spatially separated in azimuth, the head scattering would result in the increasing of monaural signal-to-mask ratio. At the same time, the binaural processing of audio information by higher hearing system is another reason of spatial unmasking. The result of the paper is significant to digital audio compression in multi-channel surround sound.Finally, the paper studied forward masking effect when two loudspeakers were spatially separated in azimuth and reproduced masker and maskee sounds respectively. The results showed that in case of spatial separation, thresholds tended to drop by a relative large amount, only very few exceptions. So, it can be conclude that the spatial separation of the sources of masker and maskee in forward masking would weaken the effect, and if high quality reproduced sounds are desired, the influence of spatial separation of sound sources should be considered when a psychoacousticl model is used in compression arithmetic.Mask effect research is a very important branch of psychoacoustics. It related to physioacoustics, neural conduction etc. The result is significant to audio information technology such as audio compression, speech enhancement and echo hiding. So, to study the mask effect deeply and widely has great significance not only to science but also to technology.
|
PDF Full Text Request