The Acoustic Beamformer is a tool that can be implemented in many environments, such as conference rooms or small lecture halls, to focus on a specific sound source in a room full of sounds. The beamformer will be an array of eight microphones, connected to an analog-to-digital converter and a PC that will collect the sound signal data from each microphone. The PC will use a custom software program to manipulate the data in order to localize a single sound source.
Acoustic Beamforming is a technique that takes inputs from multiple microphones in some kind of array (often linear) and uses the delays between the various inputs to calculate the distance from the sound source. Delay and sum beamforming applies a delay to each input and then sums the input to find the area of greatest magnitude.
Since we intend to detect speech, we have a rather wide frequency band to process. We are currently investigating options for handling speech without losing resolution. One option that we are currently pursuing is detecting formants, which are the peaks that occur when vowels are spoken. These have a less wide range of possible frequency than general speech.