Concert lovers have come to expect top-quality live music performances to match what their CD- or MP3-fed high-fidelity home sound systems deliver. However, the daunting task of ensuring a consistent aural picture that is as good as possible for all listeners wherever they may be seated, in venues ranging from large amplified concert halls to more intimate music rooms, falls on sound engineers’ shoulders. This key technician must analyze how direct and indirect sound production items interfere, decide which amplification gains are best for each loudspeaker, and possibly insert equalization and audio effects within the sound system to provide the experience musicians intend to convey. Finding a good mix of instrument configurations, microphone and loudspeaker locations, and, most importantly, sound system settings is still mostly an art.
The authors introduce a new algorithm that provides base-case mix settings, given the overall specification of an event. They approach this challenge as an offline optimization problem, using physical models of sound production and propagation and minimization techniques (such as gradient-search and genetic algorithms) to decide which configuration would be best, taking into account an ideal, target audio setting to be approximated. Multiple target mixes are handled, to accommodate the different expectations of musicians and listeners. A general outline of the technique and a description of a case study are provided. This rather technical paper will be of use to sound engineers and optimization experts, who will find here an unusual and exciting application for their techniques.