Describir: Speech Enhancement Based on LWT and Artificial Neural Network and Using MMSE Estimate of Spectral Amplitude