Spatial audio has typically been recorded from specialized microphone arrays that are too expensive and unwieldy to incorporate into today's consumer devices. Consumer devices, such as mobile devices, typically use cheap microphones with strict limitations on array geometry. Therefore, in order to capture spatial audio on such a device, we must be able to work with the given device geometry. To make spatial audio possible on a wide range of devices, a method for capturing the best spatial audio from an arbitrary array geometry is needed.In this thesis we propose several methods for capturing spatial audiofrom an arbitrary array for reproduction via headphones or binaural cross-talk cancellation. A technique for designing filters that minimize the reconstruction error of the soundfield captured at the array relative to the head related transfer function is described. Our techniques are compared with the current state of the art in spatial audio, Ambisonic recording and reproduction. Additionally, casestudies of several microphone arrangements capable of fitting a mobile device geometry are examined. Their efficacy for use in a spatial audio system is discussed.It is demonstrated that such restricted geometries are capable of capturing compelling spatial audio. In addition, it is shown that given the reconstruction techniques proposed in this thesis,performance is equal to Ambisonics when an Ambisonic array is used, and potentially superior to Ambisonics when a more flexible array isemployed.
【 预 览 】
附件列表
Files
Size
Format
View
Capturing spatial audio from arbitrary microphone arrays for binaural reproduction