Search papers, labs, and topics across Lattice.
This paper introduces a feedforward active noise control (ANC) method that leverages neural networks for time-series prediction of speech signals to effectively suppress non-stationary speech noise. By employing an adaptive filtering algorithm, the method updates a linear control filter based on both predicted and historical speech signals, addressing the limitations of traditional ANC techniques. Experimental results demonstrate significant improvements in noise reduction, showcasing the efficacy of using predicted signals in both ideal and neural network scenarios.
Neural network-driven predictions can enhance active noise control, achieving superior suppression of non-stationary speech signals.
A feedforward active noise control (ANC) method based on time-series prediction for speech signals is proposed. Although current ANC techniques are highly effective against stationary noise, suppressing highly non-stationary speech signals remains a challenging task. We propose an adaptive filtering algorithm for active speech suppression based on neural-network-based time-series prediction of future signals. The update value for the linear control filter is calculated based on the predicted signal, as well as the current and past signals. Numerical experiments indicated that the noise reduction can be improved in both cases: when using the true predicted signal and when using a signal predicted by neural networks.
