francescobrigante/SAGE
SAGE: Semantic Audio Generative Encoder. A compact Swin V2 VAE for 44.1 kHz stereo music with a CLAP-aligned latent (official code)
GitHub repository with 25 stars and 1 forks.
Language: Python
Topics: clap, fad, music-vae, semantic, music-autoencoder, latent-diffusion