Thursday, July 15, 2010

A Voice Recognition Software Demonstration

Note to those who are reading this post in an e-mail, or anywhere other than my blog actual: if you don't see the short video immediately below, this post isn't going to make a whole lot of sense to you. You might want to read this one on my actual blog, which is here.

Press play!

Note: When the VRS recognizes a word that I do not say, that word is shown in red font. The correct word, if there is one, follows in brackets, in green font. Similarly, when I say a word that the VRS misses entirely, that word is shown in brackets, in green font. (My apologies to any readers who are color blind!)

Hello everybody! This is your friendly author Krif[s]tin here, and I’m going to give you a little demonstration of my voice recognition software. Below this video, I will post the uncorrected text of this dictation. Read along, and see for yourself how well it works.

Stopover to become really like that [Here's how well the software works if you talk really really fast] about the [trains and] planes and automobiles and circus[es and] monkey[s and] never takes a [single] breath. On the other hand, here’s how well the software works if you talk a bit more slowly about planes and trains and automobiles in [and] circuses and monkeys (I believe booths [those] were all of the things I mentioned), and a brief [breathe] now and then. Here are some words that my software has learned, and almost always gets right: Graceling, Bitterblue, M*A*S*H [Nash], Graced. Here are some words that my voice recognition software [never] gets right: oh [Po]. Given [Giddon]. Fasten [Raffin]. Our door [Corridor]. Library and [Library]. I mean [Queen]. I mean [Queen]. Clean clean clean clean clean clean clean clean clean [Queen queen queen queen queen queen queen queen queen]. I just figured out that I have used the word “meme [queen]” 1003 times and [in] my current work in progress. I really can’t overstate how tired I [am] of this particular error. Sigh…

well, this has been my voice recognition software demonstration. I hope it has been helpful to those of you [who] are curious. At the bottom of this post, I will provide a little information about what software I use, but [what] its particular disadvantages are, and what software I recommend. If you have all her [arm] pain when you type, or discomfort of any kind, you really should look into voice recognition software, because arm injuries should not be taken lightly. And that’s my story! I [Bye] everybody! :-)

The first thing I'll say is that there is so much about VRS that this demonstration fails to demonstrate. Too many things to explain in this post without it getting out of hand, to be honest, but for now, I'll just say that (1) the best software has a good correction feature that allows you to correct mistakes without typing, a feature to format text by voice, a feature for training the software to recognize the words you speak, and the ability to do other things, too, like add to/subtract from/moderate the known vocabulary; and, (2) the best software allows you to do a whole lot more than dictate prose, i.e., operate various programs, have access to and control your desktop and windows, and do all sorts of computer stuff without having to type or use your mouse.

The second thing I'll say is that the absolutely only thing I use my software for is to dictate prose, so I cannot speak intelligently about the use of VRS for any other purpose -- other than what I've absorbed from listening to friends who do use it for other purposes. Also, unfortunately, my education is limited to software for English speakers. The software I know about may very well be available to speakers of other languages, but I just don't know.

The third thing I'll say is that there is one error in the dictation above that was my fault. At the beginning of the last paragraph, I should have said, "cap well" to capitalize that word. I forgot that while the software does capitalize the next word after sentence-ending punctuation (.!?), it doesn't automatically capitalize the next word after ellipses. (Also, I should have put a comma between the two instances of "really.")

(Can you believe the M*A*S*H thing? Sigh...)

Beyond all that, there's just too much to say about VRS in one blog post, so I'm going to try to wrap everything up with a little "who, what, when, where, why" section.

Who: Well, me. :-) And a WHOLE lot of other people with a WHOLE variety of needs, so many people that I really think Apple should step it up and produce some Mac-compatible software that truly competes with its far superior PC-compatible competitor. If they did, Macs could be accessible to a lot more people. But, more about that below.

What: If you're in the market for VRS for English speakers, the product you get depends entirely on whether you're a Mac user or a PC user. To the best of my knowledge, the best VRS for Mac users is MacSpeech Dictate. I'm a Mac user, and I use MacSpeech Dictate. It's... adequate for dictating prose. But if, like me, you were a PC user first, and were once able to use the far superior software Dragon NaturallySpeaking, MacSpeech Dictate will pretty much make you want to tear your hair out, guaranteed. Why? Well, I could give you numerous reasons, but here are a couple examples. (1) Dragon NaturallySpeaking allows you to teach it how to recognize your own personal pronunciation of words -- both words that are already in its vocabulary, and words you wish to add to its vocabulary. MacSpeech Dictate does not. (2) Dragon NaturallySpeaking allows you to erase words from its known vocabulary. MacSpeech Dictate does not. (If I could, I would remove the word "clean" from my software's vocabulary. That's how tired I am of the "queen" error.)

(All that being said, MacSpeech Dictate does have good word recognition. There's a lot of green and red above, but mostly just in the sections where I'm deliberately demonstrating its weaknesses. It's all the other features that are crappy. [And like I said before, I can't speak to its non-prose-dictating abilities.])

If you use a PC and are shopping for VRS, get Dragon NaturallySpeaking. No other software comes close. I know this is the case for dictation, and have friends who can vouch for it for its other abilities, as well.

When: I use my VRS when I'm transcribing my handwritten novel into my Word document. If I didn't, I would hurt myself (see below). That's the main thing I use it for, but I also use it for e-mails and blog posts whenever I'm having arm/hand/finger pain.

Where: Um, wherever I am with my computer? I try to do it where no one can hear, because I don't want people to hear me dictating my novels.

Why: I have this thing called thoracic outlet syndrome (TOS), which isn't the point of this post, so I won't get into it too much here. Basically, it's a neck/shoulder/arm/wrist/hand/fingers pain thing, + hand/finger agility thing, that I can manage about 80% of the time with daily stretches; that 17% of the time causes me discomfort, but doesn't limit what I'm able to do (you've seen the pictures); and 3% of the time makes simple motions like turning a doorknob or flipping through a pile of paper prohibitively painful and renders me incapable of typing. Copious typing is on the small list of things that causes it to flare up to that level. (You can tell it's not flaring in the video above, because if it were, I wouldn't be using my hands to talk.) This is why VRS feels like a miracle to me. I mean, I'm a professional writer. If I had to do all that typing with my fingers, I would be in disabling pain a lot more than 3% of the time (if I could even actually do that much typing, which I'm not sure I could). VRS saves my arms, which makes an enormous difference to my life. (There are so many people who need VRS for so many reasons, many who need it 100% of the time! Come on, Apple, step it up!)

I guess that's it for today. Disclaimer: There are various frustrations that you'll become familiar with if you ever decide to try VRS. It's a little like one of those unbearable people who thinks he's smart, and is always trying to correct you, but is actually dumb as nails. But I need it, and I'm so grateful for its existence.

I hope this has been illuminating!