It is possible to listen to the sound from a desired direction and isolate the noise from others by properly aligning and mixing the signals of a microphone array. Moreover, large arrays produce better sound directionality, however, they are unrealistic due to the cost and complexity of the existent hardware. We developed a new multichannel system capable of capturing the audio signals of huge microphone arrays via Light Emitting Diodes (LEDs) and a high speed video camera. Our prototype employs a single Graphics Processing Unit (GPU) to perform massive parallel processing in order to achieve real-time performance. With such a large-scale system, superdirective audio focusing in wide areas will be possible in the future.
Please click the thumbnail image to open the full-size PDF file.