Enhancing the performance of parallel network consisting of noise reduction and echo cancellation algorithms

There have been several methods to enhance the speech corrupted by echo and broadband noise. Predominantly speech enhancement can be categorized as echo cancellation and noise reduction. Several techniques were introduced to reduce various kinds of background noise and echo; some frequency techniques are based on spectral subtraction. In any mobile communication system (uplink or downlink) the corrupted speech will pass through a cascade or parallel network of noise reduction and echo cancellation modules. In this paper we describe a method to address the efficiency issues in the parallel network of noise reduction and echo cancellation modules by introducing a correction module so as to achieve performance as close as possible to the cascaded approach.