fix: broken Colab (no-UI) and UI mode after restructuring

- Create missing app/tabs/downloads/ with download_tab() function
  (GUI was crashing on launch due to missing download tab import)
- Update config.json: fix all old rvc/ paths to new engine/ paths
  (convert_path, create_dataset_path, preprocess_path, extract_path,
   create_index_path, train_path, create_reference_path)
- Fix directml.py: stale library/ path now resolves relative to package
- Fix gui.py: add missing theme=theme parameter in gr.Blocks()
- Update docs (README.md, optimizer.md, vocoder.md) to reflect new
  folder structure (services/, models/, engine/, ui/)

Author: BF667

README.md

@@ -313,19 +313,20 @@ Advanced-RVC-Inference/
 │   ├── api/
 │   │   └── cli.py              # Full CLI interface (rvc-cli)
 │   ├── configs/                 # Model configs (v1, v2, ringformer, etc.)
-│   ├── core/                    # Core utilities (UI, process, training, restart)
-│   ├── library/
+│   ├── services/                # Core utilities (process, training, restart)
+│   ├── ui/                      # UI package
+│   ├── models/
 │   │   ├── backends/            # GPU backends (CUDA, ZLUDA, DirectML, OpenCL)
-│   │   ├── algorithm/           # Model architectures and algorithms
+│   │   ├── algorithms/          # Model architectures and algorithms
 │   │   ├── generators/          # Vocoder implementations
 │   │   ├── optimizers/          # Training optimizers
 │   │   ├── predictors/          # F0 extraction algorithms
 │   │   ├── embedders/           # Speaker embedding models
 │   │   └── onnx/                # ONNX export utilities
-│   ├── rvc/
-│   │   ├── infer/               # Inference engine & audio conversion
+│   ├── engine/
+│   │   ├── inference/           # Inference engine & audio conversion
 │   │   ├── realtime/            # Real-time voice conversion
-│   │   └── train/               # Preprocessing, extraction, training
+│   │   └── training/            # Preprocessing, extraction, training
 │   ├── uvr/                     # UVR5 audio separation library
 │   └── utils/                   # Shared variables & utilities
 ├── Advanced-RVC.ipynb           # Google Colab notebook
@@ -394,7 +395,7 @@ rvc-cli info
 python -c "import torch; print(torch.cuda.get_device_name(0))"
 
 # If STFT issues occur, check that ZLUDA STFT override is loaded
-python -c "from advanced_rvc_inference.library.backends import zluda; print(f'ZLUDA: {zluda.is_available()}')"
+python -c "from advanced_rvc_inference.models.backends import zluda; print(f'ZLUDA: {zluda.is_available()}')"
 ```
 
 ### Common Dependency Issues

advanced_rvc_inference/app/gui.py

@@ -106,6 +106,7 @@ def launch(
         else:
             with gr.Blocks(
                 title=f"Advanced RVC Inference",
+                theme=theme,
                 js=js_code if client_mode else None,
             ) as app:
                 gr.HTML(

advanced_rvc_inference/app/tabs/downloads/__init__.py

@@ -1,3 +1,7 @@
-from advanced_rvc_inference.app.tabs.downloads.downloads import download_tab
+"""
+Downloads tab for Advanced RVC Inference Gradio interface.
+"""
+
+from .downloads import download_tab
 
 __all__ = ["download_tab"]