Neural Directional Filtering with Configurable Directivity Pattern at Inference

TL;DR

This paper introduces a neural directional filtering method that allows users to define spatial filtering patterns during inference, leveraging a FiLM-based architecture for flexible and generalized directivity control in audio applications.

Contribution

The work presents a novel neural network architecture with FiLM conditioning for user-defined directivity patterns, enabling flexible and generalized spatial filtering during inference.

Findings

UNDF outperforms conventional methods in experiments.

It generalizes to unseen directivity patterns.

It can approximate irregular directivity shapes.

Abstract

Spatial filtering with a desired directivity pattern is advantageous for many audio applications. In this work, we propose neural directional filtering with user-defined directivity patterns (UNDF), which enables spatial filtering based on directivity patterns that users can define during inference. To achieve this, we propose a DNN architecture that integrates feature-wise linear modulation (FiLM), allowing user-defined patterns to serve as conditioning inputs. Through analysis, we demonstrate that the FiLM-based architecture enables the UNDF to generalize to unseen user-defined patterns during interference with higher directivities, scaling variations, and different steering directions. Furthermore, we progressively refine training strategies to enhance pattern approximation and enable UNDF to approximate irregular shapes. Lastly, experimental comparisons show that UNDF outperforms conventional methods.