If you're not sure what we mean by the game "Telephone", it's the party game where you whisper something in someone's ear, then they whisper it in someone else's ear, and so on. The message gets changed over time. If you're playing this game at a party, it probably means it's a pretty lame party.
If you turned on the Closed Captions, you may notice that the transcript is significantly different from what we ended up with for each round. That's because each time a video is uploaded, the computer listens to the audio and interprets it fresh. So, even if you upload essentially the same audio, if one small thing changes in your video, the transcript will change, sometimes a lot. It's just a phenomenon associated with the technology. Our scripts from rounds 2 & 3 came from the actual transcripts that were generated when we uploaded those videos. Make sense?