WavCube: Unifying Speech Representation for Understanding and Generation via Semantic-Acoustic Joint Modeling
概要
arXiv:2605.06407v1 Announce Type: cross Abstract: Integrating speech understanding and generation is a pivotal step toward building unified speech models. However, the different representations required for these two tasks currently pose significant compatibility challenges. Typically, semantics-or…
