Unsupervised speaker adaptation based on the cosine similarity for text-independent speaker verification

Abstract