README.md

---
license: cc-by-nc-sa-4.0
tags:
- video-to-audio generation
- selective sound generation
- multimodal deep learning
---

# SelVA: Hear What Matters! Text-conditioned Selective Video-to-Audio Generation

[![arXiv](https://img.shields.io/badge/arXiv-2512.02650-brightgreen.svg?style=flat-square)](https://arxiv.org/pdf/2512.02650)  [![githubio](https://img.shields.io/badge/GitHub.io-Demo_page-blue?logo=Github&style=flat-square)](https://jnwnlee.github.io/selva-demo/) [![githubio](https://img.shields.io/badge/GitHub-Code-blue?logo=Github&style=flat-square)](https://github.com/jnwnlee/selva/)



```bash
.
├── weights/
│   ├── video_enc_sup_5.pth # text-conditioned video encoder
│   ├── generator_small_16k_sup_5.pth # v2a generator
│   ├── generator_small_44k_sup_5.pth
│   ├── generator_medium_44k_sup_5.pth
│   └── generator_large_44k_sup_5.pth
└── ext_weights/
    ├── synchformer_state_dict.pth # pretrained Synchformer (24-01-04T16-39-21)
    ├── best_netG.pt # BigVGAN vocoder
    ├── v1-16.pth # vae 16kHz
    └── v1-44.pth # vae 44kHz
```