Search papers, labs, and topics across Lattice.
1
0
3
Adversarial training doesn't have to hurt speaker verification: by explicitly modeling language, you can disentangle speaker and language characteristics without sacrificing speaker discriminability.
