| # DoodleBook — Dependencies (ZeroGPU Version) | |
| # Core | |
| gradio>=6,<7 | |
| python-dotenv | |
| spaces | |
| # Image generation | |
| diffusers>=0.28 | |
| transformers>=4.40 | |
| accelerate | |
| torch | |
| pillow | |
| sentencepiece | |
| opencv-python-headless | |
| numpy | |
| # Voice narration (VoxCPM2). MUST be pinned: unpinned, pip resolved to voxcpm | |
| # 1.5.0 whose generate()/sample-rate API differs → 4 min of garbled audio. | |
| # 2.0.3 is the version the local Modal worker uses and is known-good. | |
| voxcpm==2.0.3 | |
| # Book building | |
| fpdf2 | |
| # Utilities | |
| requests | |
| huggingface_hub | |
| soundfile | |