The Genetic Algorithms for The Optimization of Distributed Loudspeaker Systems

In some large enclosures, such as underground stations or factory halls, which involve long reverberation time, low signal-to-noise ratio, public address systems are often used to improve the speech intelligibility. Nevertheless, too many distributed loudspeaker systems suffer an inappropriate use of loudspeakers. This can lead to an imperfect uniformity of audience coverage decreasing the speech intelligibility. The use of ray tracing method associated with genetic algorithms is presented as a promising tool for the optimization of positioning and for the definition of loudspeaker number in a given room. Thus, this combination can help the designer in the conception of public address systems. The validity of this approach is illustrated in the case of choosing a best configuration from 15 possible loudspeaker locations in classroom in order to obtain an optimum speech intelligibility and to allow a smaller cost of renovation.