Improving Reconstruction Loss Based Speaker Embedding in Unsupervised and Semi-Supervised Scenarios.

Jaejin Cho, Piotr Zelasko, Jess Villalba, Najim Dehak

Browse the full ICASSP paper archive.