---
license: mit
datasets:
- SoundMind-RL/SoundMindDataset
language:
- en
base_model:
- Qwen/Qwen2.5-Omni-7B
pipeline_tag: audio-to-audio
---

# SoundMind Model 

The SoundMind Model is an audio language model (ALM) trained using SoundMind, a rule-based reinforcement learning (RL) algorithm designed to equip ALMs with advanced bimodal reasoning capabilities.

[Github](https://github.com/xid32/SoundMind) [Paper](https://arxiv.org/abs/2506.12935) [Dataset](https://huggingface.co/datasets/SoundMind-RL/SoundMindDataset) [Model](https://huggingface.co/SoundMind-RL/SoundMindModel)

## Citation

If you find our work helpful, feel free to give us a cite.

```bibtex
@article{diao2025soundmind,
  title={SoundMind: RL-Incentivized Logic Reasoning for Audio-Language Models},
  author={Diao, Xingjian and Zhang, Chunhui and Kong, Keyi and Wu, Weiyi and Ma, Chiyu and Ouyang, Zhongyu and Qing, Peijun and Vosoughi, Soroush and Gui, Jiang},
  journal={arXiv preprint arXiv:2506.12935},
  year={2025}
}
```