Work

A method for extracting an average scattering coefficient for room acoustic modeling

Year: 2025

Type: article

Abstract: This work introduces frequency-dependent Average Scattering Coefficients (ASCs) to quantify the room-averaged degree of scattering per mean free path that results from sound wave interactions with objects, elements, and impedance discontinuities. In this work, the sound field of a room is decomposed into two components: (1) specular reflections from boundary surfaces and (2) non-specular component by the room's interiors and elements. Time-dependent coherence coefficients extracted from impulse responses of furnished rooms and their empty averages are used to derive ASCs, which can be used to facilitate the transition from a specular to a non-specular component in room acoustical modeling. This study extracts ASCs in rooms with varying amounts, distributions, and absorptions of interior elements, and different source and receiver positions based on a wave-based solver. Moreover, ASCs are measured in a real furnished room and utilized to reconstruct the measured sound field with a hybrid model. The specular component is calculated from the empty room case, while the non-specular component is modeled with a stochastic technique. Results from the hybrid models show strong agreement with ground truth regarding early decay time, reverberation time, the degree of scattering, and the level of diffuseness, demonstrating the potential of ASCs for high-frequency room acoustic modeling with reduced computational resources. • Average scattering coefficients (ASCs) to evaluate total scattering effects in rooms for room acoustic modeling. • Derivation of ASCs from coherence analysis of simulated and measured sound fields in empty and furnished rooms. • Demonstration of insensitivity of ASCs to source and receiver positions in the high-frequency range. • Application of ASCs to model the transition from specular to non-specular components in hybrid models. • Reconstruction of measured sound fields using a hybrid model with measured ASCs employed.

Source: Applied Acoustics

DOI: 10.1016/j.apacoust.2025.110604

Authors: Dingding Xie, Wouter Wittebol, Qi Li, Maarten Hornikx

Institutions: Harbin Engineering University, Eindhoven University of Technology

Language: English

Topic: Speech and Audio Processing

Subfield: Signal Processing

Field: Computer Science

Domain: Physical Sciences

Open Access status: hybrid

Estimated APC paid: $3,820

Funder: China Scholarship Council