Read it Again, Sam: Using Text-to-Speech to Proof Your Novel

Brains are unreliable, not that they’d ever admit it. A perfect example can be found in the proofing of a novel. Even on the tenth or twentieth read-through of a work-in-progress, you can still find typos and missing words and the like. And why? Because the brain has had enough of your piddly story, and in an effort to get back to thinking about how that drinking bird toy works, starts glossing over your text and missing the mistakes. Fortunately, there is an answer in technology.

Popular wisdom holds that you should read your story aloud at some point so you can get a sense of how it flows, if the language sounds natural, and to catch some mistakes that might otherwise slip by. What popular wisdom doesn’t address, however, are the weird looks you get from your dog as you sit at your desk and mumble to yourself.

But more than that, reading aloud isn’t foolproof. It’s still your brain reading the words on the page, and still susceptible to the tricks and shortcuts it usually uses when reading long blocks of text. That is to say, it can skip around, it can insert words, and it can even merge two repeated words into one.

Obviously, we need something more reliable to read the story to us. And before you say have a friend read it to you, that’s just another brain. See how they stick together like that?

Text-to-Speech

I’ve used text-to-speech (TTS) since Veneer, and back then, the voices sounded like what gravel would sound like if it could talk. In those days, having a computer read your novel was almost an insult; it could make Shakespeare sound like shit.

Fast-forward to 2018 and things are better but not altogether perfect. For the longest time, it seemed like advancements in TTS were coming slowly, and there wasn’t much interest in the technology. But then came Siri and Alexa and Cortana, and suddenly TTS is a big thing again.

Recently, I went looking again for TTS software and found that some had closed and been absorbed into Amazon Polly, which I believe is some kind of TTS API that will led you add realistic speech to your apps. That’s not surprising, as Alexa sounds pretty good these days.

I didn’t find an app that made the service available in the way I need it, but luckily there are other options.

Natural Reader

I’ve used Natural Reader (naturalreaders.com) to help proof my last few books, and for the most part, it was okay. There was no way I was going to load a 120k novel into the app, but the whole “select text and press play” using the floating bar worked fine. What I needed was an app that would read every single word on the page, and this one sufficed.

That isn’t to say NR is perfect. The interface feels old and clunky. Copying text into the app gets tedious. The voices aren’t that great. They release a new version every year or so that is basically the old version. You can only install it on one computer (the website says 2, but I can’t seem to get it to work). 

They now have an online reader that you can do the same copy/paste dance with, but it costs $10 a month. I understand why companies are rushing headlong into the subscription pricing model, but I still don’t like it.

Imagine paying $9.99/mo to read a book. Only $99/yr if you pay annually.

My biggest problem with NR is that there is no pronunciation customization available like in other apps. When you write Science Fiction, you’re gonna make up some words and use lots of funny-sounding names. If the TTS app can’t pronounce your lead character’s name correctly, you’re gonna have a bad time.

So, this year, I’d had enough, and it was goodbye to Natural Reader.

Word 365

I don’t remember when Microsoft added TTS to Word, but I do remember it being terrible. So it was with great surprise the other day that I tried it out and found it sufficient. Sure, there was still no customization and I couldn’t use the voices I’d purchased for Natural Reader, but the basics were there. 

The Read Aloud button should be on the Review tab in Word. Clicking on it brings up the TTS controls in the upper right-hand corner of the document. 

You can control the reading speed and choose a voice to your liking. I find David to be agreeable, though sometimes I switch to Zira for chapters with a female narrator. Is that sexist? I don’t know.

What Word’s TTS has going for it is that it is built in. Just place the cursor, hit the play button and you’re off. It even highlights the words as it reads, so you can follow along. 

Based on that convenience alone, I prefer it to Natural Reader, and have been using it during my latest read-through. Hopefully, one day, Microsoft will hire some developers to expand this feature and add all the little niceties that would make it indispensable.

In Sum

I really enjoy this part of the revision process–listening to a computer read my book. Not only do I get to hear it read aloud, but it also signals an end in sight for the editing process. This is the back half of the course. All that’s left after this is polish and cutting and finding the inevitable embarrassing typos.

The error above made it through 5 drafts. Five. It took a computer pronouncing the word correctly to make me realize what I’d written.

So yeah, use TTS in at least one read-through so you can avoid assing your entire novel.

Share this:

Leave a Reply