Free studio vocal data for audio generation pipelines

Hi — Amphion's audio generation coverage is solid. The vocal synthesis side is where most builders hit a wall — clean, legally-cleared singing data that is AI-ready.

We dropped a free demo pack — 50 full vocal stem packs, ~150 minutes, dry WAVs. Direct download, no gate.

https://sonovox.ai/products/demo-vocal-dataset

Happy to iterate on format if something would integrate better with Amphion's pipeline.