Blind dereverberation of single channel speech signal based on harmonic structure

This paper presents a new method for dereverberation of speech signals with a single microphone. For applications such as speech recognition, reverberant speech causes serious problems when a distant microphone is used in recording. This is especially severe when the reverberation time exceeds 0.5 of a second. We propose a method which uses the fundamental frequency (F-0) of target speech as the primary feature for dereverberation. This method initially estimates F-0 and harmonic structure of the speech signal and then obtains a dereverberation operator. This operator transforms the reverberant signal to its direct signal based on an inverse filtering operation. Dereverberation is achieved with prior knowledge of neither room acoustics nor the target speech. Experimental results showed that the dereverberation operator estimated from 5240 Japanese word utterances could effectively reduce the reverberation when the reverberation time is longer than 01 of a second.