I've started messing with a programme called Omni Describer. It's a Windows Programme, and it's for getting audio descriptions of videos using AI. You can import a video in to it which you've already downloaded, or you can also paste in a link directly from YouTube. Once you've done that, you then go to the option to describe the video. Once it's finished generating the description, it'll come up with a video player to play back the video with the audio description in it. The audio description will be read out using your screen reader, using whichever voice you use normally set at the rate you normally have it at. The player has the usual options on it to play, pause, rewind and fast forward. You can also if you want export an mp3 file of the video with the description in it, and if you do that, the description will be read out by one of the Microsoft voices. There's different settings too for example how descriptive you want the descriptions to be, the speed of the voice and other things. One thing it does seem to do though is interrupt itself to go on to the next description before it's finished the one I'm on, but it was suggested to me to speed up the voice a bit, and then perhaps that might prevent it doing that so much. Oh and you can also export a txt file of all the descriptions on a video, so if you do miss anything, you can look at them. I'm loving it though so far.

You can get the programme from https://audioses.com/en/yazilimlar.php