Search papers, labs, and topics across Lattice.
4
0
8
5
ChildVox reveals the current capabilities and limitations of audio and speech models in understanding the nuanced and developing acoustic world of children.
Acoustic and phonetic NACs encode accent in fundamentally different ways, with implications for how we interpret and manipulate these representations.
Speech tokenizers, despite being crucial for multimodal LLMs, primarily capture phonetic information, creating a semantic mismatch with text-derived semantics that hinders performance.
A new multimodal dataset links brain activity, muscle activation, and articulation in speech, opening doors to understanding the causal chain of speech production.
