Update README.md

README.md

@@ -3,6 +3,64 @@
 
 ---
 
+## **0. System Setup Guide**
+
+### **0.1 Environment Preparation**
+
+**Hardware Requirements:**
+
+- NVIDIA GPU (≥16GB VRAM recommended)
+- CUDA 11.7+ and cuDNN 8.7+
+
+**Installation Steps:**
+
+```bash
+# Create conda environment
+conda create -n songedit python=3.10 -y
+conda activate songedit
+
+# Install dependencies (env.sh contents)
+pip install torch==2.0.1+cu117 --extra-index-url https://download.pytorch.org/whl/cu117
+pip install onnxruntime-gpu --extra-index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/onnxruntime-cuda-12/pypi/simple/
+pip install -r requirements.txt
+
+# Install audio processing libs
+conda install -c conda-forge ffmpeg libsndfile
+```
+
+### **0.2 Model Checkpoints**
+
+Download pretrained models from HuggingFace:
+
+```bash
+# Install huggingface_hub if needed
+pip install huggingface_hub
+
+# Download all checkpoints
+python -c "
+from huggingface_hub import snapshot_download
+snapshot_download(repo_id='badd9yang/songedit',
+                  local_dir='checkpoints')  # Optional for private repos
+"
+
+# Expected folder structure:
+checkpoints/
+├── step1/
+│   ├── separate_model.pt
+│   ├── whisper/
+│   ├── ...
+│   └── align.ckpt
+└── step2/
+    ├── whisper-small/
+    ├── model_v1.pt
+    └── model_v2.pt
+  
+```
+
+> **Note:** For manual download, get models from [HuggingFace Repo](https://huggingface.co/badd9yang/songedit/tree/main)
+
+
+
 ## **1. Core Features Overview**  
 
 ### **1.1 Song Editing Pipeline**  
@@ -167,4 +225,4 @@ We extend gratitude to the open-source community:
 
 > *"From raw audio to professional vocal production – all in one pipeline."*  
 
-[Contact Support](yangchen@hccl.ioa.ac.cn) | [GitHub Repository](https://github.com/badd9yang) | [API Reference](diffsinger.com) 
+[Contact Support](yangchen@hccl.ioa.ac.cn) | [GitHub Repository](github.com/badd9yang) | [API Reference](diffsinger.com)