Robust binaural rendering with the time-domain underdetermined multichannel inverse prefilters

Area of Science:

• Acoustics
• Signal Processing
• Audio Engineering

Background:

• Binaural audio reproduction aims to create immersive sound experiences.
• Existing systems often face limitations in robustness and sweet spot size.
• Multichannel inverse filtering is a key technique in audio rendering.

Purpose of the Study:

• To propose a robust binaural audio rendering system.
• To improve the quality and spatial realism of reproduced sound.
• To widen the optimal listening area (sweet spot).

Main Methods:

• Reformulation of the multiple-input/output inverse theorem into a multichannel model-matching problem.
• Time-domain formulation of the model-matching problem as an underdetermined system.
• Application of Tikhonov regularization to limit prefilter gains.

Main Results:

• Successful implementation on a six-element linear loudspeaker array.
• Validation through binaural rendering, crosstalk cancellation, source widening, and 5.1 virtual surround.
• Objective and subjective tests confirmed the system's efficacy.

Conclusions:

• The proposed time-domain underdetermined multichannel inverse filtering approach is effective for robust binaural audio rendering.
• The system demonstrates improvements in crosstalk cancellation, source widening, and virtual surround sound.
• This method offers a promising solution for enhanced immersive audio experiences.