Select the device rather than hard-coding CUDA

Tidy the launch call and drop stray files

Prune tags that predate the taxonomy

Credit the original authors and link the repository and paper

Fix the launch keywords and the taxonomy import

Bring the Space up to the pyharp example standard (taxonomy tags, __main__ guard, show_error, v0.4.0)

Remove unused audio decode from speech editing

Use shorter ZeroGPU reservation for speech edits

Add VoiceCraft speech editing Space