This seems to fit the exact same niche as youtube-dl, would be interesting to know what this does differently.
Functionality wise it doesn't differ at all, and it shouldn't because youtube-dl is a good media scraper. Where youtube-dl falls flat (in my opinion) is maintainability, size, external dependencies and its usage as a library. Offliberate uses Offliberty as its backend and does not contain any extraction code for specific sites whatsoever and thus remains very small (the command line interface is in fact larger than the resolver). youtube-dl currently has 1436 open issues as well as 184 pull requests on GitHub. Lots of their unit-tests also fail and take ages to complete. Offliberate uses no external dependencies other than requests and does not rely on other command line utilities such as ffmpeg for content stitching. The last difference is the emphasis on usage as a library. The CLI logic is separate from the actual resolver which can work asynchronously as well as synchronous.