Search papers, labs, and topics across Lattice.
1
0
3
18
Ditch the separate models: CAST-TTS uses a single cross-attention mechanism to control TTS timbre from both speech and text, rivaling specialized models in quality.
