Design of Multichannel FIR Filter using Gradient Descent Optimizer For Personal Audio
Typical methods to implement the personal audio systems have suggested target optimization problems, finding processes to obtain the optimum solutions and results to be used as the impulse responses of the multichannel FIR filter. As a practical attempt to generate the personal audio in the vehicles, this paper proposes use of a gradient descent optimizer to computationally find the optimum multichannel FIR filter. The gradient descent optimizer minimizes the defined cost function based on the gradient, and we will propose several examples of the cost functions induced from the target optimization problems of the previous methods. Experiments in a real environment measures the acoustic contrasts between the bright and dark zones to evaluate the method.
Authors: Seunghun Kim (CTO E&M Center, LG Electronics) and Tack-Sung Choi (CTO E&M Center, LG Electronics)