Event-Based Visual Microphone Based on Specular Reflections Off Angularly Deformed Surfaces

Area of Science:

• Optics
• Acoustics
• Computer Vision

Background:

• Conventional microphones capture sound directly.
• Visual microphones use light intensity changes to reconstruct audio.
• Existing methods have limitations in data volume and field of view.

Purpose of the Study:

• To introduce a novel passive electro-optical technique for remote audio capture.
• To demonstrate the efficacy of an event-based camera for audio reconstruction.
• To improve upon existing visual microphone technologies.

Main Methods:

• Utilizing an event-based camera to detect angular deformations on a surface caused by sound waves.
• Analyzing specular reflection changes at each pixel.
• Interpreting event stream timings as signal level-crossings.
• Reconstructing audio signals using short-time Fourier sparsity.

Main Results:

• Successfully reconstructed audio signals comparable to or exceeding prior art.
• Achieved a 25-fold expansion in the field of view.
• Reduced data volume by three orders of magnitude.
• Demonstrated capabilities in speech signal reconstruction and acousto-optical passive ranging.

Conclusions:

• The event-based visual microphone (EBVM) offers a highly efficient and effective method for passive audio capture.
• EBVM surpasses conventional and intensity-based visual microphones in performance metrics.
• This technology opens new possibilities for remote acoustic sensing.