Add model card, pipeline tag, and links to paper and code

README.md

@@ -1,3 +1,49 @@
----
-license: cc-by-nc-4.0
----
+---
+license: cc-by-nc-4.0
+pipeline_tag: audio-classification
+---
+
+# MADB (Music Aesthetics Dataset and Benchmark) Models
+
+This repository contains the trained model weights and evaluation configurations presented in the paper [MADB: A Large-Scale Music Aesthetics Dataset with Professional and Multi-Dimensional Annotations](https://huggingface.co/papers/2607.06929).
+
+MADB provides a framework for evaluating and predicting fine-grained, multi-dimensional human perceptual judgments (aesthetic assessment) of music.
+
+* **Repository / Code:** [GitHub - knownree/madb](https://github.com/knownree/madb)
+* **Dataset:** [sirui1/MADB-Dataset](https://huggingface.co/datasets/sirui1/MADB-Dataset)
+
+## Setup Environment
+
+To run the evaluations, set up the environment as detailed in the official repository:
+
+```bash
+conda create -n madb python=3.11 -y  
+conda activate madb  
+pip install torch==2.7.0 torchvision==0.22.0 torchaudio==2.7.0 --index-url https://download.pytorch.org/whl/cu118  
+pip install -r requirements.txt  
+conda install -c conda-forge ffmpeg
+```
+
+## Quick Start (with MUQ Evaluation)
+
+After downloading the model files to your local repository directory, you can perform evaluations by converting audio, extracting embeddings, and running predictions:
+
+```bash
+python code/to_wav.py --input_dir sample/sample_audio/ --output_dir muq/wav/ --mode muq  
+python code/muq_extractor.py --input_dir muq/wav/ --output_dir muq/emb/  
+python code/test.py --config code/muq_config.py --output_csv muq/result/test/test.csv
+```
+
+Please refer to the [official GitHub repository](https://github.com/knownree/madb) for complete instructions on evaluation with CLAP, Qwen, and custom training steps.
+
+## Citation
+
+If you find this work or the associated models useful, please cite:
+
+```bibtex
+@article{madb2026,
+  title={MADB: A Large-Scale Music Aesthetics Dataset with Professional and Multi-Dimensional Annotations},
+  journal={arXiv preprint arXiv:2607.06929},
+  year={2026}
+}
+```