
Tonight, for no good reason I can think of, I intend to tell you about the process of book narration recording. Is this of interest to anyone at all? I have no idea. Being dull has never stopped me before.
The first step, of course, is to set up my little makeshift studio in my closet doorway, facing out. The hanging clothes are at my back, so very little sound gets reflected from that side, and my mike is set to record only from the front. So unless a truck downshifts in the street out front, there’s probably not much noise to interfere with the ethereal music of my voice.
I have a little desk, and on it set my laptop (with the Audacity recording app opened), and my Kindle (convenient for reading from; no pages to turn). In front of me hangs my microphone on its boom, and attached to the microphone are my headphones. I also keep an insulated cup of green tea close by for refreshment and throat lubrication.
The Audacity app is free, but surprisingly sophisticated. (I wonder how they support themselves.) It shows you a screen, and as you begin recording, a graphic of the sound pattern appears in a ribbon running from left to right.
Of course, the reading never goes smoothly. You flub a word. You add a word that’s not there. You burp. Such problems must be dealt with in some manner.
There are two different ways of dealing with a reading mistake. Some narrators swear by the continuous method – you just clap your hands or click a clicker, leaving a very noticeable spike in the sound graphic, and then re-read the piece you got wrong and carry on. Later, in the editing stage, you will delete the bad stuff, and all that’s left is the good stuff, and Bob’s your uncle.
I don’t use that method, though. I use what they call “punch and roll.” There’s a built-in trick in Audacity, where you can stop the recording, click on a spot just before your mistake, hit the proper keystroke combination, and the software automatically starts playing back through your earphones about five seconds previous to the spot you clicked. You listen and along and then jump right in at the spot you marked, recording the right words (you hope), then proceed from there. Terrifying to learn (for me), but pretty slick once you get the hang of it.
Proponents of the continuous method claim that punch and roll takes you out of the rhythm and the spirit of the thing. But that’s not my experience. I can maintain my rhythm and spirit just fine.
Anyway, you keep on this way until you finish reading the chapter. Each chapter gets its own separate file.
Then comes the editing phase (for me, that usually ends up happening the following day). You go back to the beginning and listen to your work through the headphones, following along with the text in the book. More often than you expect, you find you’ve read something wrong and weren’t aware of it. In my case, getting caught up in the spirit of the moment is usually the cause.) Or maybe you made a mouth click, or breathed heavily. Such things must be cleaned up, and Audacity has ways of doing that, electronic forms of cutting and pasting.
Finally, there’s mastering. Another cause for fear and trembling, before I got comfortable with it. There’s a downloadable plug-in called ACX Check that tests your recording for three parameters: Peak volume, Volume floor, and RMS. Peak and floor are pretty self-explanatory. RMS will be explained below. Amazon Audible wants consistency in the products it publishes. So ACX Check predictably launches you on a moderately challenging series of corrections, and corrections of corrections.
Historically, the first ACX Check tells me that my Peak volume is too high. So I run the whole thing through a utility called Normalization. This utility sort of averages the highs and lows all through. Once that’s done, I run the ACX Check again, and the Peak volume will be fine. But (almost always) the RMS is now too low. (RMS is a sort of average of all the peaks and troughs. I can never remember what all the initials stand for, but the M is for “mean.”) So then I have to run the Amplify utility (there’s a formula for how to set that), and I get the RMS all tidy again. But now the Peak volume is once again too high, every single time. So (with a little prayer and fasting) I run a utility called Limiter, and in most cases all the numbers are now okay.
At that point, at least technically, the recording is acceptable for Amazon Audible. No doubt there are subjective criteria that could still disqualify the file, but from a robot’s point of view, that’s how it works.
I finished Chapter V of Troll Valley today. 20% done.
[Addendum: One hour later: On thinking it over, I realize the sequence of operations is incorrect. But I’m too tired to fix it.]