---
license: mit
---

Unconditional diffusion models used for audio inpainting.

Models:

        filename: maestro_22k_8s-750000.pt
        dataset: MAESTRO
        fs: 22.05 kHz
        segment_length: 8s
        
        filename: musicnet_44k_4s-560000.pt
        dataset: MusicNet
        fs: 44.1 kHz
        segment_length: 4s