MOSS-SoundEffect

From the project README

Process extension workspace for MOSS-SoundEffect v2.0 using the upstream runtime from OpenMOSS/MOSS-TTS and weights from OpenMOSS-Team/MOSS-SoundEffect-v2.0.

This repository is an MIT-licensed Modly integration wrapper. MOSS-SoundEffect v2.0 itself, including the upstream runtime and model weights, belongs to OpenMOSS/OpenMOSS-Team. See NOTICE and the upstream model/runtime links before redistributing weights or using the model commercially.

What this extension does Accepts a text prompt in English or Chinese. Runs the upstream MossSoundEffectPipeline. Writes a .wav artifact under Workflows/MOSS-SoundEffect/ inside the workflow workspaceDir. Emits Modly process-runner JSONL messages: progress, log, done, error.

Modly contract posture Bucket: process-extension Setup seam: root setup.py Runtime seam: root mosssoundeffectprocess.py Model ownership: Modly global model assets under models/moss-soundeffect-v2-process-extension/generate-soundeffect Output contract: done.result.filePath points to a workspace .wav audio artifact; done.result.text contains JSON metadata fallback

This extension expects Modly builds with audio as a first-class ArtifactKind. For older Modly builds that only support image, text, mesh, and scene, the process still emits JSON metadata in done.result.text and the generated .wav in done.result.filePath, but the node manifest may need to be downgraded to output: "text" until audio support is available.