| Ambisonics is a series of spatial sound reproduction system based on spatial harmonics decomposition and each order approximation of sound field.Both the accuracy of directional information reproduction and the complexity of system increase with increasing order of Ambisoncis.Based on the fact that the horizontal resolution of human hearing is higher than vertical resolution,mixed-order Ambisonics(MOA)reconstructs horizontal sound field with higher order L2D spatial harmonics,while reconstructs vertical sound field with lower order L3D spatial harmonics,and thereby reaches a compromise between the perceived performance and the complexity of system.Due to the flexibility in selecting the order combination and loudspeaker arrangement,as an important branch of Ambisoncis,MOA has gradually become one of the research hotspots in spatial sound reproduction in recent years and play an important role in various application fields,such as scientific research and civil entertainment.As a new spatial sound reproduction system appearing in the early 21st century,MOA has been researched by several research groups in the world.However,it is still being developed and only a few results were reported.Moreover,some aspects of previous researches need to be improved.First,most of the existing researches are focus on the analysis of the physical properties of the sound filed reconstructed by MOA.There are seldom investigations on psychoacoustic aspects of MOA,especially lacking of psychoacoustic analysis on localization cues and the timbre.Second,most of the previous subjective experiments focused on localization perception,rare involved in the timbre of MOA.Third,the previous researches only dealt with the recreation of far-field virtual sources in MOA,rather than the near-filed virtual sources at various distances.To address the above problems,following works on MOA have been carried out in this dissertation:Firstly,because non-uniform loudspeaker arrangements are used in MOA,which may cause instability in reproduction,the stability of some typical MOA reproduction systems is analyzed.Examples include several classic layer-wise loudspeaker arrangements and increasing the number of horizontal loudspeakers.The sensitivity of MOA reproducing sound field to the slight error of the system is qualitative evaluated by the condition number of loudspeaker direction matrix.The results indicate that for layer-wise loudspeaker arrangement,the stability of L3D/L2D-order MOA can be predicted similar to the cases of conventional L=L3D-order spatial Ambisonics.When the number of horizontal loudspeakers fulfils the requirement of spatial sampling theorem,i.e.,MH≥(2L2D+1),increasing the L2D-order and the number of horizontal loudspeakers increases the sensitivity of MOA to the slight error,but in general,the stability of MOA is still acceptable.These results provide the references for the optimization and design of MOA.Secondly,the framework for analyzing the summing virtual source localization in multichannel sound reproduction with amplitude panning(includes spatial Ambisonics and MOA)is developed.Taken the 29 loudspeaker arrangement(SCUT29)as an example,the vertical summing localization in the median plane for spatial Ambisonics is analyzed.The results indicate that spatial Ambisonics is able to create summing virtual source at different elevations in the median plane.The 3rd-order Ambisonics yields the good localization effect similar to a single real source.The succedent virtual source localization experiment validates the above results.This work can be considered as a demonstration to Wallach’s classic hypothesis about the effect of dynamic cue on vertical auditory localization.Thirdly,based on SCUT29-loudspeaker arrangement,the first MOA reproduction system in China is realized.By analyzing the error of binaural sound pressures,the design guideline of MOA is developed.For a tradeoff between the complexity of system and reasonable vertical localization performance,when the hardware allows,MOA with L3D=3 is suggested,and then the L2D(>L3D)is determined by actual requirements.The succedent virtual source localization experiment validates that based on SCUT29-loudspeaker arrangement,the 3/5-order MOA reproduction system yields reasonable stability and localization accuracy.Compared with the conventional spatial Ambisonics,MOA further improves the horizontal localization performance.Fourthly,based on Moore’s revised loudness model,the binaural loudness level spectra(BLLS)in MOA reproduction are analyzed and then the influence of the number of horizontal loudspeakers on timbre is evaluated.A subjective evaluation experiment is also carried out to validate the analysis.The results indicate that below the Shannon-Nyquist limit of spatial sampling,the BLLS of MOA match well with those of target sound field.Increasing the number of horizontal loudspeakers influence little on BLLS and then the timbre.Above the Shannon-Nyquist limit,however,the BLLS for MOA deviate from those of target sound field.The extent of deviation depends on both the direction of target sound source and the number of horizontal loudspeakers.Increasing the number of horizontal loudspeakers may increase the deviation of BLLS and then change of timbre in some cases,but reduce the deviation of BLLS in some other cases.The influence of the number of horizontal loudspeakers on BLLS and timbre reduces when virtual source departs from horizontal plane to the high or low elevation.These results validate that the deviation of BLLS can be used to quantitatively analyze the extent of timbre change.The results also provided another perspective for the MOA design.Finally,a scheme of near-field compensation mixed-order Ambisonics(NFC-MOA)for creating virtual sources at various distances is proposed.Taken the SCUT29-loudspeaker arrangement as an example,the error of binaural sound pressures in NFC-MOA reproduction is analyzed,and a distance perception experiment is carried out.The results indicate that the 3/5-order NFC-MOA is able to recreate the distance localization information(at least partly).NFC-MOA with L3D=3 exhibits a reasonable accuracy of lateral distance localization.The accuracy of distance localization is higher for horizontal virtual source than that for non-horizontal virtual source.Moreover,the performance of distance perception can be further improved by including the cue of distance-dependent loudness.These results provide a helpful guidance for the design and simplification of NFC-MOA.This dissertation explores the comprehensive perceived performance in MOA reproduction system,including directional and distance localization,as well as timbre.The results provide an important basis for further research and application of MOA. |