Search papers, labs, and topics across Lattice.
1
0
2
Reducing cpWER in multi-talker speech recognition by over 1% demonstrates that soft speaker posteriors can effectively enhance model performance without the pitfalls of hard segmentation.
