---
license: cc-by-nc-4.0
---

# DialogueSidon

Two-speaker dialogue separation model based on diffusion with a VAE-32 latent space.

## Components

| File | Description |
|---|---|
| `ssl_encoder.pt2` | w2v-BERT 2.0 backbone + latent projection heads |
| `diffusion_head.pt2` | DiffusionTransformerHead — single denoising step |
| `vae_decoder.pt2` | DAC VAE decoder: latents → 24 kHz audio |
| `metadata.json` | Latent normalisation stats, scheduler config, model dims |

## Contributors
- [Wataru Nakata](https://wataru-nakata.github.io)
- [Yuki Saito](https://sython.org)
- [Kazuki Yamauchi](https://kyamauchi1023.github.io/)
- [Emiru Tsunoo](https://scholar.google.com/citations?user=34_X2z4AAAAJ&hl=ja)
- Hiroshi Saruwatari