Sound playing data acquisition method of virtual auditory system and computer equipment

Abstract

The invention provides a sound playing data acquisition method of a virtual auditory system and computer equipment. The method comprises the following steps: acquiring a plurality of target physiological parameters of a to-be-detected object; inputting the target physiological parameters into a trained individual component feature prediction model to obtain an individual component prediction matrix of the to-be-detected object; performing data reconstruction according to a frequency direction component matrix, a first diagonal matrix and the individual component prediction matrix to obtain a head-related transfer matrix of the to-be-detected object; and performing convolution processing on the head-related transfer matrix and a sound signal to obtain sound playing data of the virtual auditory system. According to the invention, the head transfer function data with high accuracy can be obtained, so that the sound playing data of the virtual auditory system with high authenticity can be generated.

Description

technical field [0001] The present invention relates to the technical field of auditory data processing, in particular to a method and computer equipment for acquiring sound playback data of a virtual auditory system. Background technique [0002] Virtual Auditory Display (VAD) is an important part of real and immersive virtual or augmented reality. An important factor in implementing a virtual auditory system is to use personalized head-related transfer function (HRTFs) data to simulate the transmission process of sound waves from the sound source to the human ear. [0003] However, the existing methods for obtaining personalized head transfer function data have the following limitations: First, HRTF requires an appropriate dimensionality reduction method, and the existing methods encode individual domain information in the decomposition weights, ignoring the independence of individual domains, affecting the Prediction accuracy of personalized HRTF. Second, the existing m...

AI Technical Summary

Problems solved by technology

[0003] However, the existing personalized head transfer function data acquisition methods have the following limitations: First, HRTF needs an appropriate dimensionality reduction method, and the existing method encodes the individual domain information in the decomposition weight, ignores the independence of the individual domain, and affects Prediction Accuracy of Personalized HRTF

Second, existing modeling methods include multiple linear regression and nonlinear methods, in which multiple linear regression cannot represent the impact of human physiological parameters on the head-related transfer function, while nonlinear methods are implemented through neural networks, but ignore the human physiological parameters. The influence of different parts

The above limitations lead to the low practicability and utility of the virtual auditory system obtained from head-related transfer function data

Images