Room geometry inference based on spherical microphone array eigenbeam processing.

Area of Science:

• Acoustics
• Signal Processing
• Computational Geometry

Background:

• Acoustic environment parameters, like room geometry, are crucial for optimizing audio applications.
• Existing methods often rely on multiple room impulse responses, which can be complex to acquire and process.

Purpose of the Study:

• To present a novel method for three-dimensional (3D) room geometry inference.
• To utilize robust and high-resolution beamforming techniques with spherical microphone arrays.
• To process uncontrolled broadband acoustic signals directly.

Main Methods:

• Estimating directions of arrival (DOAs) and time differences of arrival (TDOAs) of direct and reflected signals using high-resolution beamforming and cross-correlation analysis.
• Addressing challenges like low signal-to-noise ratios for reflections and signal correlation.
• Combining DOA and TDOA information with geometric principles for inference.

Main Results:

• Successfully estimated DOAs and TDOAs of direct and reflected acoustic signals.
• Demonstrated accurate 3D room geometry inference using the developed method.
• Validated the technique through experimental evaluations with simulated and measured data in reverberant conditions.

Conclusions:

• The proposed method offers a robust and accurate approach to 3D room geometry inference.
• It effectively utilizes spherical microphone array signal processing for broadband acoustic signals.
• This technique has the potential to enhance various audio applications requiring precise acoustic environment knowledge.