Unsupervised Audio Source Separation via Spectrum Energy Preserved Wasserstein Learning

Separating audio mixtures into individual tracks has been a long standing challenging task. We introduce a novel unsupervised audio source separation approach based on deep adversarial learning. Specifically, our loss function adopts the Wasserstein distance which directly measures the distribution distance between the separated sources and the real sources for each individual source. Moreover, a global regularization term is added to fulfill the spectrum energy preservation property regarding separation. Unlike state-of-the-art unsupervised models which often involve deliberately devised constraints or careful model selection, our approach need little prior model specification on the data, and can be straightforwardly learned in an end-to-end fashion. We show that the proposed method performs competitively on public benchmark against state-of-the-art unsupervised methods.