Our paper “TASTE2: Text-Aligned Speech Modeling and Deployment toward Full-Duplex Voice Interaction” is now on arXiv.