To show off what realtime timbre-transfer models could do, we built a website which fetches videos from YouTube and converts them either through a strings model or choir model.