Speech Enhancement Based on Bat Algorithm (BA)

Abstract

Bat Algorithm (BA) is a recently developed population-based meta-heuristic approach (Yang 2010) which is inspired by the hunting behaviour of the bats. BA is developed by Yang (2010). As a novel feature BA is based on the echolocation behaviour of microbats. BA uses the frequency tuning technique to increase the diversity of the solutions in the population, while at the same time it uses the automatic zooming and tries to balance the exploration and exploitation during the search process by mimicking the pulse emission rate and loudness of bats when searching for a prey. This algorithm turns out to be very efficient for a diverse range of problems, and its binary version has been successfully applied to image processing and classifications. The autozooming ability in microbats is manifested in the BA as automatic adjustment from exploration to exploitation when the global optimality is approaching. This is the first algorithm of its kind in terms of balancing these two key components. As a result it proves to be a very efficient optimization technique than all other meta-heuristic algorithms. This chapter intends to present the BA as an improved approach to adaptive noise cancellation in speech enhancement. Simulation results with BA are compared with the standard PSO (Kennedy et al. 2005), accelerated PSO (Yang 2010), gravitational search algorithm (Rashedi et al. 2009) and hybrid PSOGSA (Mirjalili et al. 2010)-based speech enhancement algorithms that are dealt with in this thesis. Conclusions of the present chapter clearly establish the advantages of the new meta-heuristic BA over the other algorithms in the context of speech enhancement.