Speaker Diarisation Using 2D Self-attentive Combination of Embeddings.

Guangzhi Sun, Chao Zhang, Philip C. Woodland

Browse the full ICASSP paper archive.