Signal Enhancement in AM-FM Interference

摘要

A new approach to interference suppression is developed to enhance the audibilityof signals corrupted by amplitude-modulated (AM) and frequency-modulated (FM) tonal interference. The suppression algorithm uses a short-time, least-squares estimation of the parameters of an AM-FM model of the time-varying tonal interference. The method, developed in a sine-wave analysis/synthesis framework, can be integrated with time and frequency modifications for further signal enhancement. Suppression is applied to single and multitone synthetic and actual AM-FM interference, the latter including man-made signals (e.g., siren interference) and those that occur naturally (e.g., biologic interference). The relative advantages and disadvantages of the sine-wave framework in contrast to a short-time Fourier transform overlap-add framework are described. The enhancement techniques are robust in a large range of environments and can be designed to preserve a random noise background. Finally, it is shown that interference suppression on multichannels prior to beamforming enhances beamformer performance.