Implementation of Mask Estimation & ANN for Speech Enhancement in Non-Stationary Noise Environment

Abstract: In conventional single-channel speech enhancement, typically the noisy spectral amplitude is modified while the noisy phase is used to reconstruct the enhanced signal. It provides speech enhancement under different noisy conditions. The speech power spectrum varies greatly for different types of speech sound. The energy of voiced speech sounds is concentrated in the harmonics of the fundamental frequency while that of unvoiced sounds is, in contrast, distributed across a broad range of frequencies. To identify the presence of speech energy in a noisy speech signal we have therefore developed two detection algorithms. The first is a robust algorithm that identifies voiced speech segments and estimates their fundamental frequency. The second detects the presence of sibilants and estimates their energy distribution. The use of speech enhancement algorithm removes or reduces the presence of noise. The aim of the noise reduction algorithms is to estimate the clean speech signal from the noisy recordings in order to improve the quality and intelligibility of the enhanced signal. Due to this, it presents a method for speech enhancement using mask estimation iteratively.

Keywords: Speech Enhancement, Speech Processing, Noise Filtering, Sparse Representation etc.