In recent years, neural network-based black-box modeling of nonlinear audio effects has improved considerably. Present convolutional and recurrent models can model audio effects with long-term dynamics, but the models require many parameters, thus increasing the processing time. In this paper, we propose KLANN, a Koopman-Linearised Audio Neural Network structure that lifts a one-dimensional signal (mono audio) into a high-dimensional approximately linear state-space representation with nonlinear mapping, and then uses differentiable biquad filters to predict linearly within the lifted state-space. Results show that the proposed models match the high performance of the state-of-the-art neural models while having a more compact architecture, reducing the number of parameters by tenfold, and having interpretable components.

Feedback delay networks (FDNs) are used in audio processing and synthesis. The modal shapes of the system describe the modal excitation by input and output signals. Previously, the Ehrlich-Aberth method was used to find modes in large FDNs. Here, the method is extended to the corresponding eigenvectors indicating the modal shape. In particular, the computational complexity of the proposed analysis method does not depend on the delay-line lengths and is thus suitable for large FDNs, such as artificial reverberators. We show the relation between the compact generalized eigenvectors and the spatially extended modal shapes. We illustrate this method with an example FDN in which the suggested modal excitation control does not increase the computational cost. The modal shapes can help optimize input and output gains. This letter teaches how selecting the input and output points along the delay lines of an FDN adjusts the spectral shape of the system output.

The decaying sound field in rooms is typically described in terms of energy decay functions (EDFs). Late reverberation can deviate considerably from the ideal diffuse field, for example, in scenes with multiple connected rooms or non-uniform absorption material distributions. This paper proposes the common-slope model of late reverberation. The model can be used to describe spatial and directional late reverberation variations as linear combinations of exponential decays with fixed decay times. Its fundamental idea is to determine a set of common decay times that is representative of multiple EDFs. Consequently, all spatial and directional EDF variations are described solely with amplitude changes of the respective decaying exponentials. After deriving the common-slope model, we explore different approaches for determining the common decay times for large EDF sets, whose EDFs describe different source-receiver configurations of the same environment. Among the presented approaches, the k-means clustering of decay times is the most general. Our evaluation shows that the common-slope model introduces only a small error between the modeled and the true EDF, although the common-slope model is considerably more compact than the traditional multi-exponential model. Due to its compactness, the common-slope model yields interpretable room acoustic analysis results. The common-slope model has potential applications in all fields relying on late reverberation models, such as source separation, dereverberation, echo cancellation, and parametric spatial audio rendering.