AES Fall Online Convention

To obtain high loudness, with good bass extension while keeping distortion low, and ensure mechanical protection, one needs to control accurately the motion of the loudspeaker diaphragm. Actual solutions for nonlinear control of loudspeakers are complex, difficult to implement and to tune. They are limited in accuracy due to insufficient physical models, that do not completely capture the complexity of the loudspeaker. Furthermore, the physical model parameters are difficult to estimate.
We present here a novel approach that uses a Neural Network to map directly the diaphragm displacement to the input voltage, allowing us to invert the loudspeaker. This technique allows to control and linearize the loudspeaker without theoretical assumptions and with better accuracy than a model-based approach. It is also simpler to implement.