Room Impulse Response Interpolation from a Sparse Set of Measurements Using a Modal Architecture

Abstract

In augmented reality applications, where room geometries and material properties are not readily available, it is desirable to get a representation of the sound field in a room from a limited set of available room impulse response measurements. In this paper, we propose a novel method for 2D interpolation of room modes from a sparse set of RIR measurements that are non-uniformly sampled within a space. We first obtain the mode parameters of a measured room. Using the commonacoustical pole theory, the mode frequencies and decay rates are kept constant over space, and a unique set of mode amplitudes is obtained for each measurement location. Based on the general solution to the Helmholtz equation, these mode amplitudes are modeled as periodic functions of 2D spatial location. For low frequency room modes, the model parameters are found with sequential non-linear least squares. Results show accurate spatial interpolation of perceptually relevant low frequency modes in rooms with simple geometries having non-rigid walls.

Deep Impulse Responses: Estimating and Parameterizing Filters with Deep Networks

“…However, such interpolation methods assume that IRs undergo strictly linear transformations as the spatial locations change, which is not an accurate representation of reality. More sophisticated IR interpolation techniques that are domain-specific have been proposed, however they suffer from generalization to arbitrary measurement grids and application domains [11].…”

Deep Prior Approach for Room Impulse Response Reconstruction

“…The goal is either the interpolation or the extrapolation of RIRs related to locations different from those in which physical microphones or loudspeakers are placed. We can mainly identify two categories of sound field reconstruction solutions: parametric methods [ 11 , 12 , 13 , 14 , 15 ] and non-parametric [ 16 , 17 , 18 , 19 , 20 , 21 ] techniques. On the one hand, parametric solutions [ 11 , 12 , 13 , 14 , 15 ] rely on simplified parametric models of the sound field.…”

“…Hence, the target of these techniques is to convey an effective spatial audio perception to the user rather than a perfect reconstruction of the sound field. On the other hand, non-parametric methods [ 16 , 17 , 18 , 20 , 21 , 22 , 23 , 24 , 25 , 26 ] aim to numerically estimate the acoustic field. In this category, the sound field is typically modelled as a linear combination of solutions of the wave equation [ 27 ], e.g., plane wave [ 25 , 26 ] or spherical wave [ 22 , 23 , 24 , 28 , 29 ] expansions.…”

“…In this category, the sound field is typically modelled as a linear combination of solutions of the wave equation [ 27 ], e.g., plane wave [ 25 , 26 ] or spherical wave [ 22 , 23 , 24 , 28 , 29 ] expansions. Alternatively, RIRs are expressed through their modal expansion [ 16 , 17 ] or exploiting the equivalent source method (ESM) [ 30 ]. In general, many of these techniques employ compressed sensing principles [ 31 ] in order to estimate the data from the undersampled sound field enabling the reconstruction in the target locations.…”

“…In [ 16 ] the authors adopt the common-acoustical pole and residue model originally proposed in [ 17 ] for modeling the RIR in the frequency domain, i.e., the Room Transfer Function (RTF). The model discussed in [ 17 ] splits the RTF in two components: the acoustical poles that represent the resonance frequencies of the room and the residue that depends on the source-receiver locations.…”