For me it has to be short commentary takes, because as you put the presentation together, you need the flexibility to slide the audio along the time line to match up with the visuals.
When recording short bits of commentary (a sentence or two), you’ll probably find that you’ll make less verbal mistakes. It’s also a lot easier to slot in a new bit of commentary when you realise you said something wrong, or you decide you can improve on what you originally recorded. It happens all the time.
The down side is when you have a need to use noise suppression in your audio. Now it has to be run on all the seperate audio clips, rather than one complete audio file.
For creative freedom, it’s seperate short takes for me and I’ve made a few of these types of recordings.