Empirical signal decomposition for acoustic noise detection

Abstract

This paper introduces an adaptive noise detection method for non-stationary acoustic noisy signals. The proposed approach is based on the empirical mode decomposition (EMD) and a vector of Hurst exponent coefficients. The scheme is investigated considering real acoustic noisy signals with different non-stationarity degree and signal-to-noise ratio (SNR). The results demonstrate that the EMD-based noise detector enables a better separation between the clean and noisy signals when compared to the competing methods. It also leads to an average SNR improvement of 4.4 dB for the resulting enhanced signals.