Relative Transfer Function Inverse Regression from Low Dimensional Manifold
This work addresses the RTF inverse regression problem for room acoustics, but it is incremental as it applies an existing DNN method to this specific task without major innovations.
The paper tackled the problem of generating high-dimensional Relative Transfer Functions (RTFs) from low-dimensional source-receiver poses using a supervised Deep Neural Network (DNN) model. The result showed that the model achieved lower prediction error than the free field assumption but failed to outperform linear interpolation at small sampling distances.
In room acoustic environments, the Relative Transfer Functions (RTFs) are controlled by few underlying modes of variability. Accordingly, they are confined to a low-dimensional manifold. In this letter, we investigate a RTF inverse regression problem, the task of which is to generate the high-dimensional responses from their low-dimensional representations. The problem is addressed from a pure data-driven perspective and a supervised Deep Neural Network (DNN) model is applied to learn a mapping from the source-receiver poses (positions and orientations) to the frequency domain RTF vectors. The experiments show promising results: the model achieves lower prediction error of the RTF than the free field assumption. However, it fails to compete with the linear interpolation technique in small sampling distances.