I took all the URLs from HTTP Archive that ended ".mp3" or ".mp4", deduped by host, then made a range request to them with the following:
Accept-Encoding header (like Firefox).
Accept-Encoding: identity;q=1, *;q=0 (like Chrome).
Accept-Encoding: identity (like Safari).
Accept-Encoding: gzip, deflate, br (like Edge).
To run these tests yourself, see the repo.
Warning: These URLs are media files from the web. As such many will be unsafe for work.