Exploring Quality and Generalizability in Parameterized Neural Audio Effects

Abstract: This work expands on prior research published on modeling nonlinear time-dependent signal processing effects by means of a deep neural network with parameterized controls, with the goal of producing commercially viable, high quality audio, i.e. 44.1kHz sampling rate at 16-bit resolution. These results highlight progress in modeling these effects through architecture and optimization changes, towards increasing computational efficiency, lowering signal-to-noise ratio, and extending to a larger variety of nonlinear audio effects. Most of the presented methods provide marginal or no increase in output accuracy over the original model, with the exception of dataset manipulation. We found that limiting the audio content of the dataset provided a significant improvement in model accuracy over models trained on more general datasets.