README.md

---
license: mit
---
# Token-Based Audio Inpainting via Discrete Diffusion (AIDD)

Pretrained model weights for **AIDD**, introduced in:

**Token-Based Audio Inpainting via Discrete Diffusion**  
ICLR 2026  
https://arxiv.org/abs/2507.08333

AIDD performs audio inpainting by applying diffusion in a discrete token space, enabling semantically coherent reconstruction of missing audio segments, including long gaps of up to 750 ms.

---

## Model

The model operates on discrete audio tokens produced by a pretrained WavTokenizer and performs inpainting using a Diffusion Transformer (DiT) trained with a discrete diffusion objective. The training incorporates span-based masking to model structured missing regions and a derivative-based regularization loss that encourages smooth temporal dynamics in token embedding space. The model is designed for restoring missing segments in musical audio, including long gaps.

---

## Usage

This repository provides **model weights only**.  
For code, see the official GitHub repository:

👉 https://github.com/iftachShoham/AIDD

---

## Data & Evaluation

Trained and evaluated on **MusicNet** and **MAESTRO**, using FAD, LSD, ODG, and MOS metrics.  
See the paper for full details.

---

## Acknowledgments

Built upon  
[Discrete Diffusion Modeling by Estimating the Ratios of the Data Distribution](https://github.com/louaaron/Score-Entropy-Discrete-Diffusion.git) and  
[WavTokenizer: An Efficient Acoustic Discrete Codec Tokenizer for Audio Language Modeling](https://github.com/jishengpeng/WavTokenizer.git).  
We thank the authors for making their work publicly available.

---

## Citation

```bibtex
@article{dror2025token,
  title={Token-based Audio Inpainting via Discrete Diffusion},
  author={Dror, Tali and Shoham, Iftach and Buchris, Moshe and Gal, Oren and Permuter, Haim and Katz, Gilad and Nachmani, Eliya},
  journal={arXiv preprint arXiv:2507.08333},
  year={2025}
}