A NEW BLIND DEREVERBERATION ALGORITHM USING CHANNEL SELECTION

This paper addresses the problem of reducing the reverberation effect from speech signals, which is known as dereverberation. The main idea is to modify a dereverberation algorithm based on ideal channel selection (ICS) from an algorithm with reference to a blind algorithm. The channel selection technique performs a comparison between clean and degraded speech signals to decide the channel selectivity and a blind spectral subtraction technique (SS) was used in order to estimate the clean speech for ICS. Thus, the combination of SS and ICS generated a blind dereverberation technique named proposed approach version 1, which overcame the performance of the original ICS techniques in about 35% of the reverberation quality assessment (Q(m)). Furthermore, a different concept was adopted making the channel selection threshold varies with input speech signal instead of having a fixed value, leading to a slightly different dereverberation algorithm called proposed approach version 2, which reached Q(m) performance improvement of 38% compared to ICS.