Online Supervised Acoustic System Identification exploiting Prelearned Local Affine Subspace Models

TL;DR

This paper introduces a novel supervised acoustic system identification method that leverages prelearned local affine subspace models to improve performance in noisy environments by modeling RIR variability with low-dimensional manifolds.

Contribution

The paper proposes a new algorithm that uses prelearned affine subspace models to enhance block-online supervised acoustic system identification in noisy conditions.

Findings

Significant performance improvement over state-of-the-art methods in adverse noise scenarios.

Effective modeling of RIR variability with low-dimensional affine subspaces.

Efficient approximation of evidence for optimal subspace selection.

Abstract

In this paper we present a novel algorithm for improved block-online supervised acoustic system identification in adverse noise scenarios by exploiting prior knowledge about the space of Room Impulse Responses (RIRs). The method is based on the assumption that the variability of the unknown RIRs is controlled by only few physical parameters, describing, e.g., source position movements, and thus is confined to a low-dimensional manifold which is modelled by a union of affine subspaces. The offsets and bases of the affine subspaces are learned in advance from training data by unsupervised clustering followed by Principal Component Analysis. We suggest to denoise the parameter update of any supervised adaptive filter by projecting it onto an optimal affine subspace which is selected based on a novel computationally efficient approximation of the associated evidence. The proposed method significantly improves the system identification performance of state-of-the-art algorithms in adverse noise scenarios.