This paper proposes a second-order differential microphone array (DMA) that can continuously steer in the main lobe direction. First, a general response of the second-order DMA is derived as a linear combination of a monopole and dipoles. Nine- and seven-microphone systems are proposed to realize continuous steering of the second-order DMA. The proposed system synthesizes the steered beam pattern without requiring the phase shifter by using back-to-back cardioids. Simulation results show that the nine-microphone configuration yields a nearly constant and high white noise gain, but the seven-microphone configuration has the practical advantage that it requires fewer microphones.

During a phone conversation, loud vocal emission from the far-end to the near-end space can disturb nearby people. In this paper, the possibility of actively controlling such unwanted sound emission using a control source placed on the mobile device is investigated. Two different approaches are tested: Global control, minimizing the potential energy measured along a volumetric space surface, and local control, minimizing the squared sound pressure at a discrete point on the phone. From the test results, both approaches can reduce the unwanted sound emission by more than 6 dB in the frequency range up to 2 kHz.

Although both noise suppression and speech restoration are fundamental to speech enhancement, many Deep neural network (DNN)-based approaches tend to focus disproportionately on one, often overlooking the importance of their joint handling. In this study, we propose a dual-path architecture designed to balance noise suppression and speech restoration. The main path consists of an encoder and two specialized decoders: one dedicated to estimating the clean speech spectrum and the other to predicting a noise suppression mask. To reinforce the joint modeling of noise suppression and speech restoration, we introduce an auxiliary refinement path. This path consists of a separate encoder–decoder structure and is designed to further refine the enhanced speech by incorporating complementary information, learned independently from the main path. By using this dual-path architecture, the model better preserves fine speech details while reducing residual noise. Experimental results on the VoiceBank + DEMAND dataset show that our model surpasses conventional methods across multiple evaluation metrics in the causal setup. Specifically, it achieves a PESQ score of 3.33, reflecting improved speech quality, and a CSIG score of 4.48, indicating enhanced intelligibility. Furthermore, it demonstrates superior noise suppression, achieving an SNRseg of 10.44 and a CBAK score of 3.75.

This paper describes a technique for designing a collection of beamformers, a “beamformer bank,” that approximately produces a constant-amplitude panning law. Useful in multichannel audio recording scenarios, a point source will appear with energy above a specified sidelobe level in at most two adjacent beams, and the sum of all beam signals will approximate the source signal. A design method is described in which a specified sidelobe level determines beamwidth as a function of arrival direction and frequency, leading directly to the number and placement of beams at each frequency. Simulation results are presented verifying the proposed technique's performance.