Search papers, labs, and topics across Lattice.
ITMO University
1
0
Achieving up to 21.58% equal error rates, this method outperforms traditional approaches by effectively disentangling speaker identity from synthesis artifacts.
