Exhaust and muffler noise is a challenging problem in the transport industry. While the main purpose of the system is to reduce the intensity of the acoustic pulses originating from the engine exhaust valves, the back pressure induced by these systems must be kept to a minimum to guarantee maximum performance of the engine. Emitted noise levels have to ensure comfort of the passengers and must respect community noise regulations. In addition, the exhaust noise plays an important role in the brand image of vehicles, especially with sports car where it must be tuned to be “musical”. However, to achieve such performances, muffler and exhaust designs have become quite complex, often leading to the rise of undesired self-induced noise. Traditional purely acoustic solvers, like Boundary Element Methods (BEM), have been applied quite successfully to achieve the required acoustic tuning. However, they fail at predicting all of flow-induced noise, as well as non-linear noise dissipation mechanisms. A natural candidate for this type of problem is the use of a Lattice-Boltzmann Method (LBM) solver as a CFD tool. It has already been successfully applied and validated to quantify self-induced noise of mufflers as well as complex acoustic devices performance like acoustic liners. In this paper, a muffler baseline geometry self-induced noise is assessed using the commercial LBM solver PowerFLOW. Noise generation mechanisms are identified and design modifications are proposed to atone it. The given baseline and iterations designs noise mechanisms are analyzed and the obtained noise reductions are compared and discussed.