Audio-Material Reconstruction for Virtualized Reality Using a Probabilistic Damping Model.

Area of Science:

• Acoustics and Signal Processing
• Computational Physics
• Material Science

Background:

• Modal sound synthesis requires accurate material parameters for realistic sound generation.
• Estimating these parameters from impact sounds is challenging due to external factors like support damping and sampling inaccuracies.
• Existing methods often demand precise object geometry and impact location, increasing complexity and human effort.

Purpose of the Study:

• To develop a novel technique for estimating material damping parameters from recorded impact sounds.
• To probabilistically model and account for external factors that interfere with parameter estimation.
• To reduce the human effort and data requirements for accurate material property determination.

Main Methods:

• A probabilistic generative model was developed to represent combined effects of material damping, support damping, and sampling inaccuracies.
• Maximum likelihood estimation was employed to fit a damping model to recorded impact sound data.
• The technique was validated using a synthetic dataset and a perceptual study on object identification.

Main Results:

• The novel technique effectively estimates damping parameters by probabilistically modeling external factors.
• This method significantly reduces the need for precise object geometry and exact hit locations.
• Validation studies confirmed the technique's effectiveness in parameter estimation and object identification.

Conclusions:

• The proposed method offers a more robust and efficient approach to estimating material damping parameters from impact sounds.
• It overcomes limitations of previous methods by accounting for real-world complexities and reducing manual effort.
• This advancement has implications for realistic sound synthesis and material characterization.