Search papers, labs, and topics across Lattice.
Affiliation:
4
0
5
4
A single-tower speech codec can outperform complex dual-tower architectures, achieving low-bitrate efficiency without sacrificing semantic quality.
Uncertainty-aware methods can significantly enhance speaker verification performance by reducing EER and improving separation through a unified approach that propagates uncertainty throughout the verification pipeline.
Emotion preference models can be dramatically improved by addressing both data sparsity and model bias, leading to more accurate emotional assessments in multimodal contexts.
Current speech deepfake detection systems falter dramatically against emotionally expressive attacks, with performance dropping to near-random levels on the new AffectDF benchmark.
