Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces ParaGeo, a novel method for decomposing paralinguistic variation in speech by projecting pooled key/value representations from a frozen speech language model into a shared latent geometry. This approach allows for the disentanglement of paralinguistic attributes from linguistic content, achieving significantly higher accuracy in attribute control than permutation baselines. The findings offer a structured latent space for analyzing and manipulating speech delivery, paving the way for more fine-grained control over synthesized audio.
A shared latent geometry reveals that paralinguistic speech attributes can be precisely controlled and disentangled from linguistic content within a frozen LLM.
Speech delivery varies with both the requested paralinguistic attribute and the linguistic content. We introduce ParaGeo, a matched-content decomposition of paralinguistic variation in a frozen speech language model. Synthesized audio tokens are replayed with a fixed listening prompt; pooled key/value (K/V) representations are centered and projected into a shared low-dimensional space. Our GLM-4-Voice probe spans 80 requested controls from 12 benchmark families across eight sentences. With a globally fitted calibration basis, content-held-out centroid accuracy using this basis is 9.49% versus a 1.25% permutation baseline; same-label cross-content cosine similarity is 0.285 versus 0.017, and both conditional permutation tests yield p = 0.001. A separate ten-scenario, six-style probe reveals reproducible contrast directions across scenarios. Static, additive, and temporal interventions produce attribute-, layer-, and schedule-dependent response profiles. These results provide a shared coordinate representation for measuring paralinguistic structure and an empirical starting point for latent speech control. Code is available at https://github.com/yuhanlydia/ParaGeo.