Abstract

Microphone arrays are a vital aspect of research into the field of audio signal processing, which includes speech recognition, beamforming and source separation. The more channels a microphone array provides, the more universally usable it becomes. Since there are research projects requiring recordings with up to sixteen audio channels the need for a new sound card arose. The subject of this thesis was the development, building and testing of such a sound card. To keep the whole setup reasonably small Micro-Electro-Mechanic-System (MEMS) microphones were selected. These microphones are growing in popularity as they are very small in size and MEMS microphones with even better signal to noise ratios are becoming increasingly available. A USB 2.0 interface was chosen as the protocol between computer and sound card. This was done for easy usage of the sound card, as the USB 2.0 standard includes audio recording devices and provides plug and play functionality. The new hardware and software was developed based on an evaluation board of the company XMOS. It increases the microphone count from seven to sixteen and provides the possibility to connect the microphones with ribbon cables to the sound card. Different spatial layouts of the microphone array can therefore be easily achieved. In this thesis the functional principle of MEMS microphones and the included analog-digital-converter is also displayed. The USB 2.0 protocol is studied in detail and the xCORE XUF200 processor family from XMOS is introduced. Furthermore, the developed hardware, software and prototype is introduced and documented along with some functional tests of the sound card and microphone array prototype.