Analysis and Utilization of Reverse Mode Loudspeakers

Abstract

Loudspeakers are designed to radiate sound. However, they can also capture sounds and this microphone-like behavior is called the reverse mode. The paper investigates this backward mechanism by publishing theoretical and practical results. The analysis is based on the construction of an equivalent mechanical model, while the practical results include experiments, simulations, and measurements to further explain the properties of the reverse mode. Potential utilization perspectives of speakers in event detection scenarios are also introduced and a proof-of-concept device is proposed, which performs clap detection with a neural network trained on real-world data. These results proved that accurate event detection can be obtained. During these investigations, it was assumed that the loudspeakers were inactive, i.e. their driving sources were inactive. However, the work also outlines the analysis and explanation of the more challenging problem, called the active reverse mode, in which the acoustic events are being detected through the speaker while it is actively radiating sound.

Hardware-Software Co-development for Audio and Video Data Acquisition and Analysis

“…With Early-Stopping based on the validation accuracy the training converged after 20 epochs. The source code of the training process can be found in [37].…”