Abstract

Quality of speech captured in room degrades severely due to reverberation. This in turn degrades the performance of the applications like, automatic speech recognition, telecommunication etc. Removing/Minimizing this reverberation effect is called dereverberation. In most of the cases, Room impulse response (RIR) is unknown, dereverberation in such case is called blind de-reverberation. Several blind dereverberation techniques have been proposed. In this paper we are reviewing three papers. For each paper we have discussed its algorithm, complexity and limitation. In first review [1], we discuss mathematical model and algorithm for speech enhancement using LP residual cepstrum. In second review [2], we discussed speech enhancement using statistical model of late reverberation and in last review [3] we discussed speech enhancement using linear prediction and prediction filter.