I was asked for an exercise to help with a painful left shoulder, and I said, “Maybe try my video Good Night Shoulder Circles.” And later that night it bothered me to remember that this video is only for one side, the right shoulder.

So I thought I will quickly rewrite the original text for the other side. But as you have it, I ended up doing a complete rewrite. Which cost me half a day. At first. And then I thought how about I have it read by AI instead of me, in an old British voice, David Attenborough style.

Another full work day later, yesterday night around 11pm, I finally completed a first video — only to realise that I need to rewrite the lesson text quite some more.

The lifting of the head, that’s a movement strategy I wanted to remove from the script entirely. To lift the head would introduce a new pattern, and not finishing it will not add anything useful to the lesson. And I also considered removing the shoulder circles themselves… do they really add something useful to this lesson? But then… Shoulder Circles… that’s my lesson’s title. So I had to leave them in.

Furthermore, I needed to regenerate the audio again, sentence by sentence, and then find the right timing for pauses. This I do by lying down and practicing along myself. A rather tedious process. So that’s what I’m going to be busy with today, and unfortunately maybe even longer.

Working with AI generated audio takes a lot more time than I have anticipated. But what surprises me most is this: Why is writing a lesson SO MUCH MORE WORK than just speaking it free from my heart?