The Improvement For Ambisonic System Optimization

Ambisonic system is based on the theory that soundfield can be decomposed or reconstructed by a series of spherical harmonics. Using spherical harmonics to sample the original soundfield and get the Ambisonic components is called the Ambisonic encoding process. According to the orientation of different loudspeaker, and getting the best set of coefficients to combine with different Ambisonic components to get the playback signal for the corresponding loudspeaker, is called the Ambisonic decoding process. The encoding process and decoding process of Ambisonic system are seperate completely and no affect each other. This is the reason that Ambisonic system has very high flexibility for system configuration and easy to achieve.There are two decoding way to get the combined coefficients for Ambisonic system. One way is the theoretical calculation by getting the invert matrix which containing the spherical harmonic components of all loudspeakers with different orientations, but the quality of the decoder would be affected if the loudspeaker layout is not regular enough. The other way is using the vector synthesis theory which was introduced by Gerzon and was based on the psychoacoustic researches about virtual sound imaged construction by multi sources, and this method need to use heuristic searching program with a set of fitness functions to get the best optimized solution by comparing the low frequency and high frequency quality evaluation vector rV and rE with the original sound source. A method which called the Virtual Microphone Polar Controlling method for vector based Ambisonic system optimization improvement was introduced in this paper. That is, using a set of fitness functions, the polar pattern of the virtual microphone which synthesized by virtual spherical harmonic microphone array with the decoding coefficients can be controlled, and the lobes outside of the target coverage of this synthesized virtual microphone which will capture the undesirable sound wave on these directions and cause the sound direction confusion during playback can then be minimized. It solves the nonlinear optimization problem and reduced the computation time efficiently. It leads the optimization program to get the"in-phase"decoder and it is especially effective for the high order Ambisonic system optimization with irregular loudspeaker layout. Some Ambisonic system such as the 4th order ITU layout Ambisonic system were optimized by this method and the decoding results were also given in this paper.The dodecahedron layout 3th order ambisonic system which optimized by virtual microphone polar pattern controlling method was discussed in detail in this paper. After set up the system, some acoustic parameters according to ISO3382 were measured in the center point of the Ambisonic system. Although the parameter LF and IACC were not good enough, the other acoustic parameters were very similar with the original sound field. Some listening tests were also performed in after. The result is that the system is very robust for virtual sound image reproduction and no affected by head rotation mostly, and the system has good localization quality up to 2000Hz but started to get worse at 4000Hz and higher. The result of the listening test is that the subjects can identify the different position by hearing in the same hall correctly. The evaluation value of hearing perception compared the 3th order Ambisonic system with real hall is between slight different and rather similar, with the condition that the target sources are came from the computer model of real hall.