# Hugging Face Spaces Deployment Checklist

## ✅ Files Ready

- [x] `app.py` - Main Gradio interface (HF Spaces compatible)
- [x] `requirements.txt` - All Python dependencies
- [x] `README.md` - Space metadata and documentation
- [x] `.gitignore` - Git ignore rules
- [x] `HF_DEPLOYMENT.md` - Detailed deployment guide

## 📋 Before Deploying

### Model Files
- [ ] Verify `Models/` directory contains trained model
- [ ] Verify `Preprocessing/wav2vec_embeddings/` contains all `.pt` files
- [ ] Verify `Utility/example_wavs/` contains all speaker reference audio files
- [ ] Consider file sizes - models >100MB should use Git LFS

### Code Files
- [ ] Verify `InferenceInterfaces/` contains all necessary modules
- [ ] Verify `Preprocessing/` contains all text processing modules
- [ ] Verify all imports in `app.py` are available
- [ ] Test locally with: `python app.py`

### Documentation
- [ ] Update `README.md` with your contact info (if different)
- [ ] Add link to academic paper (already included)
- [ ] Add any additional acknowledgments

## 🚀 Deployment Steps

### 1. Create Hugging Face Space
```bash
# Go to: https://huggingface.co/new-space
# - Name: austrian-dialect-tts (or custom name)
# - SDK: Gradio
# - Visibility: Public (or Private)
# - License: Choose (recommended: CC-BY-4.0)
```

### 2. Set Up Local Git LFS
```bash
git lfs install
```

### 3. Clone Your Space
```bash
git clone https://huggingface.co/spaces/YOUR_USERNAME/YOUR_SPACE_NAME
cd YOUR_SPACE_NAME
```

### 4. Copy Project Files
```bash
# Copy all necessary directories and files
cp -r /path/to/InferenceInterfaces .
cp -r /path/to/Preprocessing .
cp -r /path/to/Utility .
cp -r /path/to/Models .
cp app.py requirements.txt README.md .gitignore .
```

### 5. Track Large Files with Git LFS
```bash
git lfs track "*.pt"
git lfs track "*.wav"
git lfs track "*.pkl"
git add .gitattributes
```

### 6. Commit and Push
```bash
git add .
git commit -m "Add Austrian Dialect TTS application"
git push
```

### 7. Monitor Deployment
- Check Hugging Face Spaces UI for build status
- Wait for "Building" → "Running"
- Check logs if there are any errors

## 🔧 Testing After Deployment

1. Wait for app to finish building (usually 2-5 minutes)
2. Click the generated app link
3. Test with example text:
   - "Der Nordwind und die Sonne stritten sich"
   - "Grüß Gott, wie geht es dir?"
4. Try different speaker and dialect combinations

## ⚠️ Common Issues & Solutions

### Issue: Module not found errors
```
ModuleNotFoundError: No module named 'InferenceInterfaces'
```
**Solution**: Ensure all directories are in the repo root, not in subdirectories

### Issue: File too large errors
```
File exceeds GitHub's file size limit (100 MB)
```
**Solution**: Use Git LFS for files > 100MB

### Issue: Out of memory errors
**Solution**: May need GPU tier for inference. Upgrade Space to GPU:
- Settings → Hardware → Select GPU (T4 or A100)

### Issue: Slow inference
**Solution**: Deploy with GPU for faster synthesis (2-5 sec vs 10-30 sec)

## 📊 Performance Expectations

### With CPU (Free Tier)
- First inference: ~15-20 seconds (model loading)
- Subsequent: ~10-15 seconds per sentence

### With GPU (Paid Tier, ~$0.60/hour)
- First inference: ~5-10 seconds
- Subsequent: ~2-3 seconds per sentence

## 🔐 Important Security Notes

- ✅ No API keys needed in code
- ✅ Text input is temporary (not stored)
- ✅ Audio output is temporary (not stored)
- ⚠️ Use HTTPS (automatic on HF Spaces)
- ⚠️ Avoid hardcoding credentials

## 📝 After Successful Deployment

1. Share your Space link
2. Document any customizations made
3. Monitor usage and performance
4. Update README with any additional info
5. Respond to community comments/issues

## 🆘 Need Help?

- Hugging Face Spaces Docs: https://huggingface.co/docs/hub/spaces
- Gradio Docs: https://gradio.app/docs/
- GitHub Issues: Check the original repository
- Email: lorenz.gutscher@ofai.at

---

**Status**: Ready for deployment ✨
**Last Updated**: 2026-08-20