Folks I've wasted a ton of time on narration. Have a system now which is viable.
Not all narration in this version has been improved, you'll still find rock-bottom quality and latest-iteration quality mixed together.
Basically I sound-proofed my office (which still continues for sake of future captures), and started using a digital audio filter to help screen out what noise is left, am capturing with a Yeti behind a mic screen and in front of a sound absorber. Bumpy materials on walls. A very different capture environment and workflow, and as my wife accurately describes the room, a big ugly mess.
I also ran out of time, and now am having trouble finding enough time to just wrap up this "NASA" chapter. It is a finish line which keeps getting closer but I keep moving slower. So kinda frustrating, but what-ya-gonna-do, I'll get there ASAP.
I'm still just trying to clean this up and get it online for public to see. Sorry no thoughtful summary here, just making sure you folk can check latest iteration. Is here mostly just to share difference in narration scattered throughout.
UPDATE 2016-04-06
Bloody hell.
Ok, so a new technical problem just popped up.
https://forums.adobe.com/message/8656633
I'm recaptured good narration covering most-but-not-all p01. It has been very satisfying to finally tear out some of that terrible tinny narration and drop in what finally sounds like proper narration.
I was doing that until PPro no longer let me cut/copy/paste my narration capture. I mean it sits there as a series of long WAV clips, but I can't position them appropriately in the TR2016 timeline.
After asking for help on Adobe's forums, I realized that this was happening to the audio captures on which I'd applied an audio plug-in. It was the audio plug-in called "iZotope TX 5 Dialogue De-noise" which, in combination with Yeti mic and foam-on-walls, is why current narration sounds so much better than previous.
I've contacted iZotope, have not heard back.
It appears that a bug in "Denoise" is screwing up PPro when too many instances of the plug-in are in use. And it isn't a total-count, it is rather, some sort of internal counter that increments for each fresh instance... if I cut/paste, cut/paste, cut/paste then the total number does not go up, but the identifier for each instance still goes up.
And it has gone up past whatever buggy limit exists in that plug-in.
This is probably an unusual scenario. I have re-recorded my narration a couple times since using this plug-in. And I did enough manipulation of those earlier takes that I'm sure I pushed that internal counter up more than most. Professional narrators probably give fairly long and accurately timed delivery, where-as I fix my own timing by editing it.
In a sense, this is a show-stopper. I can not complete p01 using the plug-in.
So, while I wait for iZotope to get back to me... and ultimately probably bug-fix and issue a new update of their plug-in... I'm going to remove the plug-in as needed to continue narrating and editing.
If they don't have an update by time I'm ready to issue TR2016 p01, then it is a question of how bad is the PC noise, and is it worth waiting for a fix?
A bit of a crappy situation because it makes the editing process sub-optimal... the best workflow would be for Denoise to be on all the narration now and to edit with it. Rather than me Denoising every single tiny clip of narration later on if/when the fix arrives.
I've also been working on later chapters, due to bottlenecks on a silent house for capture. Some interesting stuff to show for next update, although still very rough edit.