Search papers, labs, and topics across Lattice.
Affiliation:
6
0
6
Anchoring sub-grid star formation and supernova feedback to invariant physical scales rather than numerical cell sizes cuts cross-resolution variance in simulated stellar mass down to ~20% across an 8脳 grid refinement.
Natural-language instructions can now handle both zero-shot speech synthesis and surgical acoustic editing within a single unified model, operating at 4-step distilled inference speeds without classifier-free guidance.
Proxy Policy Steering (PPS), an inference-time adaptation method that resolves the challenge of eliciting task-specific behavior from limited demonstrations without degrading their broad capabilities by training two lightweight proxy policies whose calibrated velocity-space difference steers the frozen base sampler.
EmoTra-TTS achieves up to 87% improvement in emotional transition quality, redefining how TTS systems can convey nuanced human emotions in speech.
Leading models fall short in emotional intelligence, with EmoS achieving 83.8% accuracy and nearing human-level performance.
VocalRender achieves a remarkable $0.42$ improvement in naturalness over the strongest baseline, revolutionizing singing voice synthesis for real-world composition.
