Research On The Extraction Of HRIRs Basis Vectors In Spatial Hearing Model

Virtual Reality technology builds a virtual environment with high-tech instrument, and in which participators can obtain the same feelings as in reality. Virtual Auditory Space is to add the 3D virtual sound simultaneously with the vision in virtual environment, and make the sound of virtual environment approach that of the real sound field.Human's hearing feelings not only include the level, tone and timbre, but also include the sense of sound location, called the spatial hearing ability. Long-term researches have been conducted in the fields of hearing physiology, hearing psychology and acoustics, and make much progress in spatial hearing. The researches show that Head-Related Impulse Response (HRIR) and Head-Related Transfer Function (HRTF) are the most important aspects in the study of spatial hearing. In this paper, some methods of signal and information processing are employed to the study of the HRIR's features and some researches on spatial hearing modeling. The content of this paper mainly includes the following parts:1. Approach a noise-removing method based on singularity detection with wavelet transform according to the properties of HRTF. And this method can remove noise effectively while keeping the singularity information of HRTF;2. Analyze the HRIR's pretreatment methods, and present the best method based on the measurement data of CIPIC Interface Laboratory at the University of California at Davis;3. Use Principal Component Analysis to extract the spatial features of HRIRs, implement the data compression, and construct the spatial hearing model on continuous sphere space with linear interpolation;4. Use Genetic Algorithm to select the basis vectors of HRIRs, and then reconstruct the HRIRs on measurement space with the corresponding weights;5. Use Neural Network technology to realized Principal Component Analysis.