A Synthesis Model for a Moving Sound Source Based on Beamforming

Moving sound sources are essential parts of outdoor environmental auralization. The moving vehicles, e.g. cars and trains, can be modeled by separable sound sources, which are represented by their locations and signals. A measurement-based synthesis model is developed to characterize moving sound sources regarding the two representations in this paper. This model utilizes beamforming to obtain the locations and signals of moving sound sources. In this paper, the target sound source is a periodic signal, and a spiral microphone array is used in the model. First of all, the Doppler effect in the recording is eliminated by the time domain de-Dopplerization technique. Furthermore, delay and sum beamforming is modified for moving sound sources to process the de-Dopplerized signals. Subsequently, the source positions are acquired by scanning the reconstruction plane. Thus, with steering the array to the corresponding positions, the beamforming outputs can be regarded as the potential signal s of the sources. Additionally, the model is evaluated by errors of localization, frequency and level with varying parameters, i.e. steering position, steering window length and source speed. Finally, validation is conducted on the model by a pass-by measurement with the periodic signal from a loudspeaker installed on a car.