What I do is read straight through the piece, no matter how long. As I am reading, if I catch an error I have made, I go back to the logical point, maybe a word, maybe a phrase or maybe an entire sentence. I then continue on is this way until the end of the piece. That way, my voice is very close to the same when I make those corrections. I then go back and listen to it again from the beginning, trying to fine the mistakes I missed in the original reading. When I find one, I use the same correction method as above. I go to the end of the recording, read the error correctly, then cut and past the new reading into the place of the error. Again,since it is in a very close time frame, my voice is usually close enough. After that, I upload it to LV.
I have found that it usually takes me about three times the length of the recording to get this process done. So a 20 minute recording will take about an hour to complete, an hour recording, about three hours. I find that doing it in one sitting allows me to keep my ever changing voice closer to the same throughout the recording.
Anyway, that is the way I have found that works best for me.
Good luck to all, and do whatever works for you.
![Very Happy :D](./images/smilies/icon_biggrin.gif)
Wayne