Enhancement of Speech Using Combined Spectral Subtraction and Wavelet Packet Transform

Wavelet Packet Transforms (WPT) is a particular linear combinations of wavelet, now becoming an efficient tool for signal analysis. Compared with the normal wavelet analysis, it has special abilities to achieve higher discrimination by analyzing the higher frequency domains of a signal. The frequency domains divided by the wavelet packet can be easily selected and classified according to the characteristics of the analyzed signal. So the wavelet packet is more suitable than wavelet in signal analysis and has much wider applications such as signal and image compression, denoising and speech coding. The WPT generalizes the time frequency analysis of the conventional wavelet transform (Coifman et al 1994). WPT decomposes both the approximation and the detail components of multiple levels. This results in more components representing the signal and provides more flexibility, which makes other types of processing possible. Nevertheless, this increased flexibility is achieved at the expenses of a more complex denoising system and consequently a higher cost (Jiang et al 2006).