README.md

---
title: WhAM
emoji: 🐋
colorFrom: blue
colorTo: indigo
sdk: docker
pinned: false
hardware: a10g-small
---

# WhAM: a Whale Acoustics Model

WhAM is a transformer-based audio-to-audio model designed to synthesize and analyze sperm whale codas. It uses masked acoustic token modeling to capture the unique temporal and spectral features of whale communication.

For full technical details, installation instructions, and training scripts, please visit the official repository at [https://github.com/Project-CETI/wham].

## Citation

If you use this code, model, or data in your research, please cite our [NeurIPS 2025](https://openreview.net/pdf?id=IL1wvzOgqD) publication:

```bibtex
@inproceedings{wham2025,
  title={Towards A Translative Model of Sperm Whale Vocalization},
  author={Orr Paradise, Pranav Muralikrishnan, Liangyuan Chen, Hugo Flores Garcia, Bryan Pardo, Roee Diamant, David F. Gruber, Shane Gero, Shafi Goldwasser},
  booktitle={Advances in Neural Information Processing Systems 39: Annual Conference
                  on Neural Information Processing Systems 2025, NeurIPS 2025, San Diego, CA, USA},
  year={2025}
}