I assumed it was an attempt to mask videos from automatic copyright scanners. Like, the software couldn’t recognize the video as being copyrighted material if the sound was very different from the original audio.