Fig. 3From: Small-parallel exemplar-based voice conversion in noisy environments using affine non-negative matrix factorizationActivity matrices for parallel utterancesBack to article page