Coding of spatial sound fields

点击翻译按钮获取中文摘要

The realization of virtual acoustic environments (VAEs) is a current trend in spatial audio research. VAEs are increasingly used in a wide range of applications, like in movie and music productions, computer games, as well as in medical science. With the processing power of up-to-date PC's and digital signal processing techniques it is possible to render complex virtual worlds in high grades of detail. Especially in three-dimensional applications, this leads to enormous amounts of data which are impractical for real-time signal processing, storage and transmission. In order to reduce the data, the main objective of this research proposal is a parametric coding approach for spherical sound fields. The focus of the experiments is on the binaural auralization of spherical microphone array data for room simulation applications. For coding, the data is evaluated perceptually in listening experiments and further related to the physical properties of the sound field. This way, various perceptual and room acoustical parameters are extracted, which are important for room acoustical quality description. They provide a basis for the parametric representation of the sound field. In the encoder, the multi-channel audio signals are down-mixed to a mono representation which is transmitted to the decoder, together with the parameters as side information. Here, the original scene is reconstructed. In the experiments, various array configurations, rooms, test signals, as well as realistic measurement conditions, like spatial aliasing or microphone noise, are taken into account.