Search papers, labs, and topics across Lattice.
1
0
3
6
Autoregressive generative models, previously unsuitable for real-time target speaker extraction, can now achieve offline-level performance in streaming scenarios thanks to a novel chunk-wise splicing technique.
