Tips on voice recognition software
by Gordon Graham, That White Paper Guy
I seldom feel like a pioneer, but in one way I am.
I've been using voice recognition software for more than 15 years now.
I've used it to write articles, chapters of books, hundreds of e-mails, and of course, major chunks of white papers.
Over the years, I've seen voice recognition (VR) software get better and better both for Windows and the Mac. Today it's superb on both.
Here are some tips I've learned during that time.
 |
VR tip #1: Now you can try VR for less than ever. Basic VR is built into Windows 7, so you can try it on a new PC for free. And now there's a Home version of Dragon Naturally Speaking for Windows selling for about $50. At prices like that, it's not much of a risk. |
VR tip #2: But be ready for pay for best results.
The VR in Windows 7 is great for navigating the desktop, controlling software, and doing consumer tasks like updating Facebook.
But nearly every reviewer agrees: For professional dictation, you'll get better results from Dragon Naturally Speaking from Nuance.
The same company makes Dragon Dictate for the Mac (after acquiring MacSpeech in 2010).
I've used both Windows and Mac versions for years, and they're both superb.
It costs about $150 for Dragon Naturally Speaking Premium with a headset included. Dragon Dictate for the Mac costs about the same.
You don't have to save many hours to earn a payback on that kind of investment. And believe me, you will save hours with it.
For geeks only: The history behind the Dragon software is quite intriguing, since it was almost lost to the world due to corporate shenanigans. You can read about that in a neat Wired article from 2003.
Of special interest: When the original architects started on Dragon in 1971, they were looking for a big problem they could likely solve in 30 to 50 years?!
They picked computer dictation at natural speaking rates running on inexpensive machines. Now, 40+ years later, this problem really HAS been solved.
VR tip #3: Get a good microphone.
VR software is a lot more forgiving than it used to be, but a quality microphone is still vital.
If you don't get a bundled headset or have an iPhone to use for a microphone, expect to pay at least $50 for a decent noise-canceling microphone.
Nuance supplies a certified headset if you buy the packaged software (not the download) and offers a list of compatible hardware here.
VR tip #4: Forget mobile apps, for now.
Yes, there is Siri and a stripped-down version of Dragon on the iPhone. Yes, there is voice recognition in android. And sure, there are some voice recognition apps for the iPad.
All these are fine for recording a quick note or asking Google for the nearest gas station.
They may even give you a glimpse of the amazing possibilities of VR technology.
But seriously, will you ever write a white paper on your iPhone? Or even your iPad? I didn't think so.
These mobile devices are intended more for viewing media than for creating it. And they lack the processing power that a VR engine demands.
For now at least, production-quality voice recognition lives on the desktop.
VR tip #5: Start with low-risk projects like e-mail.
To get started, try dictating answers to some e-mails.
You'll probably be amazed how well it works and how much time you save.
I still find dictating e-mails a real blast that saves me many minutes every day.
Let's face it, all of us can talk much faster than we type. And now that the computer can keep up, we can dictate answers to routine e-mails in our natural speaking voices.
VR tip #6: Don't worry about disrupting others.
By the way, if you're working in a cubicle, don't worry about disrupting your office-mates.
Dictating to your computer isn't any louder than talking on the phone.
You may be saying things like, "Hi again comma new paragraph here is the latest draft of your white paper period..." but hey, if they think that's weird, that's their problem.
VR tip #7: Learn your software.
I'm not a power user of Dragon by any stretch, so I know there's much more I could be doing with the software.
But you'll want to learn a few basics like these:
 |
|
- Scratch that: deletes the last few words you dictated
- Search Google for [whatever]: searches for [whatever]
- Scroll up, scroll down: scrolls through a webpage
There are dozens more commands you can use to edit, move around your files, and control your computer. Frankly, I don't bother.
VR tip #8: Use VR instead of retyping sources.
I've had white papers with 30 footnotes, where I had to type in a sentence or two from each source.
What a pain, keeping all that straight on my desk while I re-typed the text.
Then I realized there was a better way. Now I just go to each source and read in the choice bits I want to include at the appropriate spot in my draft, plus all the details for that footnote.
Meanwhile my hands are free to sort sift through my notes and file away all my sources in good order.
VR tip #9: Use VR for first drafts, not wordsmithing
Most of all, I use dictation to get a torrent of words into the computer as fast as possible.
An added bonus is that when I reread material that I dictated, I see it with fresh eyes, almost as though it was written by someone else.
That helps keep me fresh on a major project.
Then I polish and rework that text with a traditional mouse and keyboard.
For me, this is a fine way to operate. No one says you have to use your voice for everything.
VR tip #10: Think before you speak.
VR holds up a mirror to our speaking patterns. If you're a rambling speaker who goes "um... ah... you know..." every few seconds, you may not like what you see.
But if using VR helps improve your basic communication style, it's all for the best.
When I'm dictating, I generally pause between sentences to gather my thoughts on what to say next. Then I try to speak in more or less complete phrases and sentences to minimize the amount of rework to do later.
Sometimes I even say the same sentence twice, the second time in more fluent prose, knowing I can quickly delete the first version as I polish.
VR tip #11: Don't imagine you can write a white paper just by talking.
Your VR software will likely shave hours off your first draft. But after that, you still have some serious rewriting and polishing to do.
No software for any price can replace the need for rigorous thinking, persuasive arguments, and solid proof. You will still have to provide all that.
VR tip #12: You can't get both sides of a phonecall.
 |
If only some inexpensive software could record and transcribe BOTH sides of a telephone interview. Then we could sit back, yack on the phone, and get crisp, perfect transcriptions handed to us. |
Writers often ask me if that's possible. I have to tell them we're not there yet, for several reasons:
- Everyone's voice is different. It's a big challenge for software to handle two different voices in rapid-fire succession.
- Accents are a problem. Imagine a Brit speaking with a Jamaican or a Korean. Even human listeners are challenged by different accents. More so for software.
- The sound quality on phone calls can be poor. Especially if you're using a mobile device, a VOIP service like Skype, or a speakerphone.
- The sound level on each side of a phonecall is generally different. That creates another technical challenge.
Some day we will probably achieve perfect transcription of multi-voice communications.
But by then, I'm not sure transcriptions will be so exciting to anyone.
Bonus tip: Have your computer read to you
Voice recognition is voice-to-text. But a text-to-voice option is included in your PC, and it makes a good tool for white paper writing.
Under Windows 7, press the Windows key+U. With Narrator running, put your cursor in your draft and press Insert+F6 to read the current paragraph, or Insert+F7 to read the current page.
On the Mac, highlight the text you want to read and press Control+T.
You can adjust the default voices, and even get additional voices and software specifically for text-to-voice.
Sure, they all sound more-or-less robotic, but this is still a great way to check how your draft sounds when read aloud.
Whenever I do this, I'm alerted to numerous little fixups that can make the phrasing smoother and the sentences crisper.
By the way, did you know that traditional "proofreading" was a two-person operation, where one read from the original manuscript, while the other looked at a typeset proof? The reader used a pencil to indicate punctuation: one tap for a comma, two taps for a full stop.
In a strange way, that brings us full circle to today, when we have to say "comma" and "period" when dictating to a machine.
Why wait for spring? Try it today!
I believe the few hundred dollars I've spent on software and microphones over the years has been repaid many, many times over in fast and easier composing.
I don't use VR every day for every project. But I love having it available as a "secret weapon."
Whenever I need to pump out a quick first draft, or even deal with a mini-avalanche of e-mail, I'll likely pop on my headset and start talking, not typing, into my computer.
Try it yourself, and see.
What do you think? Got a comment or question?
Click here to join the discussion on LinkedIn.
