A Privacy-preserving Unsupervised Speaker Disentanglement Method for Depression Detection from Speech (short paper).