Near-field microphone array design for a hands-free system in a vehicle by using the nash genetic algorithm

The near-field beamforming method is investigated in this study for a replacement of current hands-free system of a cell phone. The goal of this array system is to design a beamformer covering 250Hz to 8kHz of voice frequency band in conjunction with constant beamwidth technique. Though the acoustical array is effective in a reverberant field, its use is limited at low frequencies because arrays longer than 1 meter are difficult to install inside a vehicle. Furthermore, it is especially impractical to use circular coordinates inside a vehicle because the sound source, a human mouth, remains a fixed distance from and moves parallel to the surface of the vehicle’s headliner. For the delay-and-sum beamformer in a near field, it is difficult to derive the complex coefficients, namely, the amplitude weight and time delay, by analytical methods. This study first describes a sound-field analysis using one popular method, the finite-element method. By obtaining the global information, the data reveal the characteristic of the interior sound field near the installation space for a microphone array. Second, a numerical optimization method for designing a fixed microphone array in the near field is investigated that uses the Nash genetic algorithm (Nash GA), which was originally used in electromagnetism problems. The Nash GA determines the optimal geometry of an element alignment for the inside of a vehicle and then generates the complex coefficients at the first stage. Unlike in a far-field array, the phase response in a near-field array plays a key role. Therefore, a second stage is added to filter out coefficients with which the responses are cancelled out in the main lobe. Third, based on the results from the algorithm, the array is implemented with two PCI-4472s, which are 8-channel simultaneous analog sampling devices. The results, measured in an anechoic chamber, correspond to the results of a computer simulation. The array performance satisfies the requirements for hands-free mobile telephony where the microphone array is installed in a vehicle’s headliner. To investigate the signal-to-noise ratio, the data is also measured in a reverberant field, like a classroom.