Search papers, labs, and topics across Lattice.
University of Palermo, Georgia Institute of Technology, Norwegian University of Science and Technology
3
0
5
33
A single-layer speech enhancement model outperforms naive architectures and achieves competitive quality with a significant speedup through progressive knowledge distillation.
Unbury speech from cinematic sound effects by teaching the model to "listen" for how words are formed.
LLMs can effectively aggregate diverse speech quality metrics, even outperforming specialized models when labeled data is scarce.
