Snejana Pleshkova, Viktor Hristov, Alexander Bekiarski
Microphone Arrays, Sound Source Localization, Real Time Processing, Parralell Programming, GPU, CUDA
Microphone arrays are the basic audio sensors delivering the appropriate information from which is possible to determine the direction of sound source arrival. There are a lot of methods and algorithms proposed for effective microphone array information processing like time delay estimation, cross correlation, beamforming, etc. All of these methods need the extensive calculations difficult to realize in real time for practical applications in area of mobile robot speakers tracking, noise cancellation, video conferencing, etc. Some implementations with digital signal processors (DSP), field programmable gate arrays (FPGA) and other high speed circuits are proposed to decrease time of microphone array information processing. The goal in this article is to apply popular in computers GPU (Graphic Processing Units) and their appropriate programming platform CUDA (Computer Unified Device Architecture) for the purposes of real time microphone array information processing.
Cite this paper
Snejana Pleshkova, Viktor Hristov, Alexander Bekiarski. (2016) Real Time Processing of Microphone Array Information Applying GPU Unit and CUDA Platform. International Journal of Signal Processing, 1, 186-191