1 paper · 1 filter
Gael Le Lan, Bowen Shi, Zhaoheng Ni +9
We introduce MelodyFlow, an efficient text-controllable high-fidelity music generation and editing model. It operates on continuous latent representations from a low frame rate 48…