I am speaking – not typing – this sentence into my computer, which is a new experience. For a long time I have been curious about hands-free computing with voice control. The gold standard in this area, of course, is Dragon NaturallySpeaking – so I finally contacted Nuance Software and received a free sample.
Note: My review sample was Dragon NaturallySpeaking 12 Premium. Potential users of other editions, or Dragon Dictate for Mac, please remember that your system might be different from what I describe here.
Setup and training
Inside Dragon’s blue-and-green box, you will find five things. The first and most important item is a piece of bubble wrap, thoughtfully included to give your hands something to do instead of typing. (For the rest of this review, imagine the sound of me popping the bubbles between phrases.) Then there’s the Dragon DVD itself, a quad-fold quick reference card, a headset “Y” adapter, and a freebie headset. “Freebie” is not exactly used as a term of endearment here. This headset is definitely in the lightweight division, and its mic boom makes a clicking sound like a nylon tie when you adjust it. The best I can say about the headset is, it works – and it will do if nothing else is available.
Setting up Dragon is easy. There is the usual short installation process, followed by ten minutes or so of training. This training is mutual – Dragon trains you on how to use the software, but the main point is for you to train it so it will recognize your particular speaking style. This involves reading out loud from one of several selections made available by Dragon. I chose to deliver the first few stanzas from John F. Kennedy’s inauguration speech. After this process is complete, Dragon offers to scan your email and documents in order to learn your vocabulary (a step I skipped because it doesn’t recognize Mozilla Thunderbird and OpenDocument files) and then it’s ready to use.
After setting up my profile and starting Dragon for the first time, I fired up OpenOffice Writer and dictated a few paragraphs from a nearby book. I was immediately blown away – out of my ergonomically-adjustable office chair, through the window of my office, and into the basketball goal outside. Okay, I’m exaggerating, but you get the idea – I was impressed. I had heard that Dragon was phenomenally accurate, but there’s nothing quite like seeing your own normally-spoken words transcribed with near-perfection, as if by the invisible hand of an efficient stenographer.
I don’t mean to suggest that Dragon always transcribes everything perfectly, or that it can somehow magically understand the name your cousin picked for her chinchilla out of a Bulgarian telephone book. Its mistakes – and new words not yet in the vocabulary such as names – can be corrected by voice: You simply say, “Correct blah” (where blah is the word Dragon got wrong) and then choose the suggested alternative, or just spell the desired word – again, by voice if you like. There’s also a handy vocabulary editor that will allow you to add special custom words and whole phrases, with their pronunciations spelled out.
Is this the dawn of hands-free computing?
At the beginning of this review, I mentioned “hands-free computing,” the cherished dream of those whose hands have suffered under computer tyranny. If we could use our computers without touching them, wouldn’t that be the ultimate ergonomic experience? We would never have to worry about Carpal Tunnel Syndrome again, and we could work from any position we found comfortable – even while exercising, provided we didn’t run out of breath.
In reality, total voice computing would likely bring its own set of problems. For instance, I know of at least one user who got laryngitis, and ultimately a sinus infection, from overdoing it with voice control. Carpal Tongue, anyone?
Even if hands-free computing were the theoretical end-all of ergonomics, Dragon is still far from good enough to make it practical for everyone. In fairness, complete hands-free control would require something more like a mind-reading program than voice transcription software. No matter how smart they may look to us, computers are still far too dumb to just “know” exactly what we want and do it.
For example, working in OpenOffice Writer I can give a few simple commands – “New paragraph, capitalize Jason, delete that” – and Dragon will understand what I want. But suppose I add a simple diagram from OpenOffice Draw. I tell Dragon, “Click on the blue oval and drag it about two inches to the right.” Dragon responds “Please say that again,” or else brings up its bewildered “dictation box” and starts inanely transcribing my instructions. And who can blame it? The program has no idea what a “blue oval” is, “drag it” is not a valid command, and “about two inches” means nothing.
There is, in fact, a way to accomplish the above task with Dragon – but it isn’t exactly easy. It goes something like this: “Mouse grid.” (A numbered grid overlays the screen.) “Four.” (Referring to Section 4 of the grid, which gets the pointer into the general vicinity of where you want it.) “Move mouse left. Move mouse left faster.” (Dragon’s default notion of “faster” comes from the fable of the tortoise and the hare, without the hare.) “Stop. Move mouse right. Stop. Move mouse down. Move mouse down faster. Stop. Click. Drag mouse right….”
I’m sure you get the idea. Theoretically, you can do anything with Dragon that you can do with a keyboard and mouse – but in reality, most people wouldn’t want to use a computer that way unless they had no other option due to a disability. For those who find themselves in that situation, the Pro Edition makes things a bit easier by including a macro system so the user can pre-automate common tasks. For the rest of us, the dream of transparent hands-free computing is still just that – a dream.
How Dragon changes my writing
“With time, I will get used to composing out loud.” These are some of the first words in Dragon’s standard training. I said them because they were in the script, but I’m not at all sure I believe the statement myself. If I’m trying to write something “polished” – like this blog post, for instance – I find it very hard to work through dictation. This is largely because I write with what might be called the “trial and error” method. Each sentence, each phrase – each word – gets deleted and replaced multiple times, often as fast as I can type it, until I feel satisfied that it “reads” well. For me, trying to go over all the options mentally ahead of time is like trying to play chess without a board.
From my perspective, Dragon is ideal for quick-and-dirty writing tasks where speed is paramount and style doesn’t matter. Two good examples are instant messaging and note-taking – particularly drudgery note-taking, like when I have to summarize a phone conversation in the company CRM, or document a programming solution I already spent two hours creating. The software excels at taking down a sloppy draft of sloppy thoughts, transcribing raw nouns and verbs very quickly into something readable.
If I try to use Dragon for my higher-grade writing tasks, two things will happen. First, the writing will take a long time. This is ironic, because the software can transcribe roughly twice as fast as most people – including me – can type, even accounting for errors. Yet it’s true. The transcription software makes me hesitate, pause, make sure what I’m about to say sounds reasonable before committing it to the microphone.
Also, the software makes verbosity so easy that it’s almost the path of least resistance. A voice-written article of mine tends to be long and rambling rather than short and to the point. For fairly obvious reasons, it also tends to require much more editing than a typewritten one.
The tale of a user who lived happily ever after
For balance, I should now share the story of someone who did get used to composing out loud, and likes it. This person happens to be my mom, a writer, who was starting to experience some computer-related hand stress issues from typing. I installed her copy of Dragon the same time I installed mine and – unlike me – she has been using it regularly ever since. Her dictated writing is now indistinguishable in quality from her standard style. (As you may have guessed, she doesn’t use the trial-and-error method like me.)
Mom absolutely loves the software, as it lets her write as much as she wants to – up to a thousand words a session – with no physical repercussions. She has even found that she can dictate successfully in the middle of a noisy room, and in a voice so quiet that no one else can make it out. If there ever was a poster child for Dragon NaturallySpeaking, it’s Mom.
A big, (resource-) hungry dragon is this one
If you are reading this blog post using Internet Explorer 7 on a Windows XP computer you bought during the first term of George W. Bush, don’t go out and get Dragon 12. Instead, go out and get a new computer. I’ll wait.
Got the new computer? Great. I hope you ordered extra RAM, because Dragon NaturallySpeaking has an insatiable appetite for that commodity. The system calls for at least four gigabytes. My laptop has six, but when other programs are running, Dragon can still bring it to its knees. It’ll chew up all the processor power you can feed it as well – in some cases while you’re not even dictating! Trying to run the thing on an older computer is something I wouldn’t recommend unless you have no alternative.
If all else fails, read the instructions
It is no secret that I am a ready-fire-aim kind of guy. I tend to jump right into something new without “wasting time” reading the documentation. (Which, ironically, ends up wasting far more time as I wander around lost with the map in my pocket.) Reading the docs is always a good idea, but with Dragon it can mean the difference between frustration and pleasure, flailing and productivity. So don’t be like me – take your time when going through the tutorial, read the quick reference card, pay attention to the tips in the Dragon sidebar, and above all, use the help file.
In summary, Dragon is great if you do a lot of writing – but don’t expect to throw away your mouse
If you do much instant messaging or make piles of notes on the computer, you should probably be using Dragon. If you write for a living – and you don’t favor my patented trial-and-error approach – Dragon could save a lot of miles on your hands and fingers. Just don’t expect your computer to become an electronic Jeeves, carrying out verbal instructions spoken between sips of a smoothie. Total hands-free computing is still out of reach, at least for now, but mind-blowing software transcription is here today. It’s called Dragon NaturallySpeaking.
Disclosure: As mentioned above, I received a complimentary copy of Dragon NaturallySpeaking in exchange for my honest review. After the review, I also signed up to be a Nuance affiliate, which means that I receive a commission on Dragon products sold through my links.