Debug monkey-patch with error logging

Use /respite/ prefix with working Gradio mount structure

Inject Respite routes into Gradio server via launch monkey-patch

Restore working structure with FastAPI + Gradio mount

Minimal baseline to confirm RUNNING state

Use /respite/ prefix to avoid Gradio proxy conflict

Patch demo.launch to inject FastAPI routes into Gradio server

Switch to Docker SDK for full ASGI control

Add API routes directly to Gradio FastAPI app

Switch Space SDK to python for direct FastAPI serving

Expose mounted FastAPI app for Spaces

Let Spaces own the application server

Keep API startup independent of model prewarming

Make Space startup logging ASCII safe

Avoid non-ASCII startup UI output

Mount Respite API with Gradio UI

Add Respite API discovery metadata

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload folder using huggingface_hub

Fix sample rate to 44.1kHz and run generation on CPU (GPU decorator only on startup check and load_model)

Fix model names: use small-music and small-sfx instead of full HF IDs

Add SFX model support with model selector

Fix ZeroGPU compatibility and auth

Update app.py

Upload 3 files