Search papers, labs, and topics across Lattice.
4
0
5
2
CoReLoop enables additional refinement in an already-trained SSL-based detector while preserving the detector's original first-pass prediction by training only lightweight refinement modules and loop-specific low-rank adapters on the original data.
Achieving a remarkable 1.454% EER, Teffic-Audio outshines all public competitors in the challenging landscape of speech deepfake detection.
Paralinguistic cues can be effectively harnessed in dialogue systems, leading to a 175% improvement in safety response accuracy without compromising overall model performance.
SLMs that seem safe with text inputs can completely fail when the same content is spoken, revealing a critical "speech grounding gap" in current models.
