Deep Audio Prior
This addresses audio processing problems for researchers and practitioners by enabling data-free methods, though it builds on prior work like Deep Image Prior.
The paper tackles audio processing tasks without training data by introducing Deep Audio Prior (DAP), which uses a randomly-initialized neural network with audio-specific priors, achieving superior results on tasks like blind source separation and audio editing compared to previous work.
Deep convolutional neural networks are known to specialize in distilling compact and robust prior from a large amount of data. We are interested in applying deep networks in the absence of training dataset. In this paper, we introduce deep audio prior (DAP) which leverages the structure of a network and the temporal information in a single audio file. Specifically, we demonstrate that a randomly-initialized neural network can be used with carefully designed audio prior to tackle challenging audio problems such as universal blind source separation, interactive audio editing, audio texture synthesis, and audio co-separation. To understand the robustness of the deep audio prior, we construct a benchmark dataset \emph{Universal-150} for universal sound source separation with a diverse set of sources. We show superior audio results than previous work on both qualitative and quantitative evaluations. We also perform thorough ablation study to validate our design choices.