Speech enhancement using transient components in frequency domain

This paper presents a method for enhancing intelligibility of speech signals using the transient components in the time domain. The Wavelet transform is used for extracting the transient components, and a ramp function with a damped motion models the behavior of the transient components. The wavelet transform of transient component represents a wavelet atom and is calculated by superposition of these atoms. Enhancement of speech signal is achieved by amplifying the location of extracted transient components in frequency domain. In this approach, there is no need to pre-process the tonal components, and the obtained results show an enough complexity of this method with respect to the other enhancement methods.