diff --git a/README.md b/README.md
new file mode 100644
index 0000000..94e3c92
--- /dev/null
+++ b/README.md
@@ -0,0 +1,218 @@
+# ComfyUI-FireRedTTS2
+
+A ComfyUI integration for FireRedTTS2, providing high-quality dialogue and monologue speech synthesis capabilities.
+
+## Features
+
+- **Dialogue Generation**: Multi-speaker conversation audio generation
+- **Monologue Generation**: Single-speaker narrative audio generation  
+- **Voice Cloning**: Zero-shot voice cloning functionality
+- **Multi-language Support**: Chinese, English, Japanese, Korean, French, German, Russian
+- **Automatic Model Download**: Models download automatically on first use
+- **Device Adaptive**: Automatically selects optimal device (CUDA/MPS/CPU)
+
+## Installation
+
+### Method 1: ComfyUI Manager (Recommended)
+
+1. open [ComfyUI Manager]
+2. Search for "ComfyUI-FireRedTTS" in ComfyUI Manager
+3. Click Install
+
+### Method 2: Manual Installation
+
+1. Clone this repository to your ComfyUI custom nodes directory:
+```bash
+cd ComfyUI/custom_nodes
+git clone https://github.com/1038lab/ComfyUI-FireRedTTS.git
+```
+
+2. Install dependencies:
+```bash
+cd ComfyUI-FireRedTTS
+pip install -r requirements.txt
+```
+
+3. Restart ComfyUI
+
+### Model Download
+
+On first use, the system will automatically download the FireRedTTS2 model from Hugging Face:
+- Model source: `FireRedTeam/FireRedTTS2`
+- Storage location: `ComfyUI\models\TTS\FireRedTTS2`
+- Download size: ~2GB
+
+A progress bar will show during download. Once complete, the model is cached for future use.
+
+## Nodes
+
+### FireRedTTS2 Dialogue Node
+
+Generates multi-speaker dialogue audio.
+
+**Inputs:**
+- `text_list` (STRING): Dialogue text with speaker tags
+- `temperature` (FLOAT): Controls generation randomness (0.1-2.0, default: 0.9)
+- `topk` (INT): Controls sampling range (1-100, default: 30)
+- `prompt_wav_list` (STRING, optional): Reference audio file paths (multi-line)
+- `prompt_text_list` (STRING, optional): Reference text list (multi-line)
+
+**Outputs:**
+- `audio` (AUDIO): Generated dialogue audio
+- `sample_rate` (INT): Audio sample rate (24000Hz)
+
+### FireRedTTS2 Monologue Node
+
+Generates single-speaker monologue audio.
+
+**Inputs:**
+- `text` (STRING): Input text content
+- `temperature` (FLOAT): Temperature parameter (0.1-2.0, default: 0.75)
+- `topk` (INT): TopK parameter (1-100, default: 20)
+- `prompt_wav` (STRING, optional): Reference audio file path
+- `prompt_text` (STRING, optional): Reference text content
+
+**Outputs:**
+- `audio` (AUDIO): Generated monologue audio
+- `sample_rate` (INT): Audio sample rate (24000Hz)
+
+## Usage
+
+### Speaker Tag Format
+
+Use square brackets to mark different speakers in dialogue text:
+
+```
+[S1]Hello, what a nice day![S2]Yes, perfect for a walk.[S1]Shall we go to the park?[S2]Great idea!
+```
+
+**Supported speaker tags:**
+- `[S1]` - Speaker 1
+- `[S2]` - Speaker 2  
+- `[S3]` - Speaker 3
+- `[S4]` - Speaker 4
+
+### Reference Audio Format
+
+**Reference audio list** (one file path per line):
+```
+/path/to/speaker1_sample.wav
+/path/to/speaker2_sample.wav
+```
+
+**Reference text list** (one line per speaker):
+```
+[S1]This is reference text for speaker 1
+[S2]This is reference text for speaker 2
+```
+
+## Examples
+
+### Basic Dialogue Generation
+
+1. Add "FireRedTTS2 Dialogue" node
+2. Input in `text_list`:
+   ```
+   [S1]Welcome to our podcast![S2]Today we'll discuss AI development.[S1]That's a fascinating topic indeed.
+   ```
+3. Adjust `temperature` and `topk` parameters
+4. Connect audio output to preview or save node
+
+### Voice Cloning Dialogue
+
+1. Prepare reference audio files
+2. Input audio paths in `prompt_wav_list`:
+   ```
+   /path/to/voice1.wav
+   /path/to/voice2.wav
+   ```
+3. Input corresponding text in `prompt_text_list`:
+   ```
+   [S1]This is a voice sample for speaker one
+   [S2]This is a voice sample for speaker two
+   ```
+
+### Monologue Generation
+
+1. Add "FireRedTTS2 Monologue" node
+2. Input long text content in `text` field
+3. Optionally provide `prompt_wav` and `prompt_text` for voice cloning
+4. Adjust parameters and generate audio
+
+## Parameter Guide
+
+### Temperature
+- **Low (0.1-0.5)**: More stable, consistent speech
+- **Medium (0.6-1.0)**: Balanced stability and naturalness
+- **High (1.1-2.0)**: More variation and expressiveness, may be unstable
+
+### TopK
+- **Low (1-20)**: Conservative sampling, more stable speech
+- **Medium (21-50)**: Balanced choice
+- **High (51-100)**: More diverse sampling, increased variation
+
+## Troubleshooting
+
+### Common Issues
+
+**Q: Model download fails**
+A: Check network connection and Hugging Face access. Try using proxy or mirror sites.
+
+**Q: CUDA out of memory**
+A: 
+- Reduce input text length
+- Lower batch size
+- Use CPU mode by setting `device="cpu"` in code
+
+**Q: Poor audio quality**
+A:
+- Check input text format is correct
+- Adjust temperature parameter (recommended 0.7-1.0)
+- Ensure reference audio quality is good (if using voice cloning)
+
+**Q: Speaker tags not working**
+A:
+- Ensure correct tag format: `[S1]`, `[S2]`, etc.
+- Check for extra spaces around tags
+- Confirm text contains corresponding speaker tags
+
+**Q: Node loading fails**
+A:
+- Check dependencies are properly installed
+- Verify ComfyUI version compatibility
+- Check console for error messages
+
+### Performance Optimization
+
+**Memory Optimization:**
+- Long texts are automatically split for processing
+- Model instances are cached and reused
+- Recommended single text length: under 500 characters
+
+**Speed Optimization:**
+- First use requires model download, subsequent uses are faster
+- GPU acceleration significantly improves generation speed
+- Batch processing multiple short texts is more efficient than single long text
+
+### System Requirements
+
+**Minimum:**
+- Python 3.8+
+- 4GB RAM
+- 2GB storage space (for models)
+
+**Recommended:**
+- Python 3.9+
+- 8GB+ RAM
+- NVIDIA GPU (4GB+ VRAM)
+- SSD storage
+
+## Support
+
+If you encounter issues, please check:
+1. Dependencies are fully installed
+2. Models downloaded correctly
+3. Input format meets requirements
+4. System resources are sufficient
+
+For more technical details, refer to the project source code and FireRedTTS2 official documentation.
\ No newline at end of file
diff --git a/requirements.txt b/requirements.txt
new file mode 100644
index 0000000..ce47e35
--- /dev/null
+++ b/requirements.txt
@@ -0,0 +1,15 @@
+# Core PyTorch dependencies
+torch>=2.0.0
+torchaudio>=2.0.0
+
+# FireRedTTS2 model dependencies
+torchtune>=0.1.0
+transformers>=4.30.0
+huggingface_hub>=0.16.0
+
+# Audio processing and utilities
+einops>=0.6.0
+tqdm>=4.64.0
+
+# Additional dependencies for model functionality
+numpy>=1.21.0
\ No newline at end of file