Combined Temporal and Spectral Processing Methods for Speech Enhancement

Speech signals from the uncontrolled environments may contain degradation components along with the required speech components. The degradation components include background noise, reverberation and speech from other speakers. The degraded speech gives poor performance in automatic speech processing tasks like speech recognition and speaker recognition and is also uncomfortable for human listening [1]. The degraded speech therefore needs to be processed for the enhancement of speech components. Several methods have been proposed in the literature for this purpose, majority them can be grouped into spectral processing and temporal processing methods. In spectral processing methods, the degraded speech is processed in the transform domain, where as, in temporal processing methods, the processing is done in the time domain, for enhancing the speech components. Each of them have their own merits and demerits. These two approaches may be effectively combined by exploiting their merits and aiming to minimize the demerits. This may lead to speech enhancement methods which are more effective and robust compared to only spectral or temporal processing. Exploration of the same is the motivation for this research work. The methods proposed in this work are therefore termed as Combined Temporal and Spectral Processing (TSP) methods for speech enhancement.