Scalable microphone array system optimized for streaming inverse acoustics

Sound source localization by using large microphone arrays commonly requires expensive hardware and an impractical large amount of cables and connectors. Our aim is to develop a compact and low-cost acquisition system that is expandable to hundreds of channels. Our design consists of a ring network of multiple sensor modules, each equipped with 16 input channels. One of the sensor modules interfaces with a computer through USB and functions as a master. Standard UTP cables are used to link the modules together for data transport and power supply. This principle reduces both weight and cost of cables. To increase streaming data rates, multiple ring networks can be connected to a computer to enable further expansion of the array. As the raw data stream is proportional to the number of elements in the array, data rates can exceed the maximum bandwidth of standard computer interfaces and storage devices. To overcome this problem, each module has a buffer for up to 45 seconds of raw data storage. Additionally, each module carries a digital signal processor (DSP) for optional signal processing. In the current configuration we preprocess each channel in order to reduce data rates and enable inverse acoustics as a streaming application.