I thought if one is using 10secs (I'm unsure if there is a real number or duration) of any video, song, or literature it is not 'reproducing' or distributing IP or copyright, but Fair Use, and therefore not against a civil or criminal law.
Fair Use is about HOW a copyrighted work is used, not simply HOW MUCH of it is used.
If the source material is readily identifiable, and it is not clearly apparent that the re-user is engaging in a protected action like academic study, critical review, or parody, then the odds are pretty good that in the eyes of the law it will be considered a derivative work, and a copyright violation if not properly licensed.
Adding the video aspect of this work actually makes it MORE likely that the source material will be identifiable. You probably wouldn't be able to tell from 2-second audio-only snippet that a drum pattern was originally performed by Bernard Purdie, but when the audio is accompanied by the video footage of him actually playing it in one of his instructional videos, it gets a lot more identifiable.