To show off what our models could do, we built a website which fetches videos from YouTube and converts them through one of our 2 best performing timbre-transfer models.