Search papers, labs, and topics across Lattice.
2
0
6
5
The resulting 82M-parameter model, Wayu-Paxa-TTS-Edge, enables on-device Thai TTS without reference audio and achieves the lowest pause-placement error and intra-word pause rates among the three systems.
Forget text-first: SODA models show that scaling native audio foundation models with interleaved semantic, acoustic, and text tokens unlocks powerful audio generation and cross-modal capabilities.
