1 paper · 1 filter
David Perera, François Derrida, Théo Mariotte +2
Training speech separation models in the supervised setting raises a permutation problem: finding the best assignation between the model predictions and the ground truth separated…