节点文献

多通道立体声虚拟重建技术研究

The Study of Virtual Multichannel Surround Sound Reproduction Technology

【作者】 王超

【导师】 支琤;

【作者基本信息】 上海交通大学 , 信号与信息处理, 2009, 硕士

【摘要】 随着数字信号处理和计算机技术的不断发展,“虚拟现实”技术已经成为人们广泛研究的课题。其中的虚拟声重建技术是近年来发展比较迅速的一门边缘交叉科学,涉及听觉心理学、听觉生理学、人工智能和高性能计算机系统等多个研究领域。虚拟声重建技术广泛应用到多媒体技术、各种计算机3D游戏以及虚拟战场等科学研究和消费电子领域。本文研究的主要目的是对多通路立体声进行相互之间的虚拟重建。在声像定位研究的基础上,把5.1通路立体声转换为2通路立体声(以下简称为5->2),实现2通路耳机对5.1通路声源的有效播放,增强耳机播放的环绕效果;把2通路立体声转换为5.1通路立体声(以下简称为2->5),实现5.1通路播放设备对2通路声源的有效播放,达到一定的环绕效果;最后把两者相结合,实现2通路立体声到5.1通路立体声再到2通路立体声的转换(以下简称为2->5->2),增强了原始2通路立体声的环绕感与包围感。本文的5->2转换算法主要基于声像定位技术,介绍了基于HRTF的声像定位原理,通过对常用定位方法的分析,利用虚拟头部转动的方法,改进了水平面声像定位的方法,有效提高了水平面上声像定位的准确率,并且提出了三维空间的虚拟声信号处理方法,即用三维空间的HRTF数据进行信号的馈给,进行了三维声像定位角度的理论分析和实验研究,在转换以后听者可以感受到更好的三维立体感与包围感。本文的2->5转换算法主要基于信号去相关处理,对已有的去相关方法进行分析,提出了基于24个临界频带的子带随机延迟法,即把声频信号分为24个子带信号,分别对左右声道各个子带信号进行随机延迟处理,提高了算法的去相关性能,使得生成的后方环绕声不影响前方的声像定位,并且提供一种三维立体感与包围感。本文将5->2和2->5转换算法相结合,把2->5转换以后的结果,再依据5->2的转换方法,生成新的2通路立体声。在2->5转换中产生了后方环绕声信号,并且在5->2转换中引入一个3D处理模块,同原始两声道信号相比,转换后的2声道信号将具有更好的环绕感和立体感。本文的研究最终在PC平台上实现了一个多通道立体声的相互虚拟重建系统。实验结果表明,该系统能够对不同通道数的音频信号进行相互之间的有效转换,可以提供给不同的音频播放系统。尤其是经过2->5->2处理以后的普通2声道信号,用耳机重放具有更好的立体感与环绕感。

【Abstract】 With the development of digital signal process and computer science,“virtual reality”technology has become a very hot researching point. In recent years,“virtual sound reproduction”, as one branch of“virtual reality”, has developed a lot. It includes Psychoacoustics, Physioacoustic, Artificial Intelligence and High Performance Computer System fields.“Virtual sound reproduction”technology has been widely applied in multi-media, 3D-game and virtual battlefield and so on.The purpose of this research is to realize the transforming between Multichannel Surround Sounds with different channel numbers. Based on the research of sound image location, transform 5.1 channel stereo sound to 2 channel stereo sound (use 5->2 for short), using 2 channel earphone to listen to the 5.1 channel stereo sound, enhance the surround effect of earphone playing. And transform 2 channel stereo sound to 5.1 channel stereo sound (use 2->5 for short), using 5.1 channel system to play the 2 channel stereo sound, with surround sound effect. And finally by combining these two methods, realize the transform of 2 channel signal to 5.1 channel signal and then to 2 channel again (use 2->5->2 for short), the surround sound effect is greatly enhanced. The 5->2 transform method is mainly based on the sound image locating technology. Presented HRTF based sound image locating technology, and by analyzing the general locating method, introduced a novel horizontal sound image locating method, which increased the precision of horizontal sound image locating. And put forward a three-dimensional virtual sound process method, which using 3D HRTF data when processing the sound signal. And analyzed three-dimensional sound image locating theory, did related experiments. After introducing the three-dimensional method, can get a better surround sound effect. The 2->5 transform method is mainly based on the signal de-correlating arithmetic, analyzing the existed method, put forward a 24 critical sub-bands random delay method. Through this method, enhanced the performance of de-correlate process, lowed the relativity between the front sound and the back surround sound, and brought forward a three-dimensional stereo effect.We combined the 5->2 and 2->5 transform methods. Using the 2->5 transform method, transformed a 2 channel source stereo sound to 5.1 channel stereo sound, and using the 5->2 transform method, transformed the new 5.1 channel stereo sound to a new 2 channel stereo sound. In the 2->5 transform process, produced new surround sound; and in the 5->2 transform process, introducing the 3D process module, produced surround sound with 3D effect. Compared to the original 2 channel sound, the new 2 channel sound has a better stereo surround effect.We realized the Virtual Multichannel Surround Sound Reproduction system on a personal computer. The experiment results showed that this system can realize the transforming between Multichannel Surround Sounds for different play system. Especially for the 2->5->2 transform method, we can get a new 2 channel stereo sound with better surround effect compared to the original sound.

节点文献中: