Using Speech Recognition Software

I noted that the month of August was my best month for writing, possibly ever. I averaged nearly 2000 words a day, combining writing for my blog with some book editing duties for GM Igor Smirnov. Despite that, I perhaps used less time than my previous months. How is that possible? Simply, I never typed a single word. I used speech recognition software.

It’s supposed to make writing this easy, and it sorta does that. Sorta.

I wrote previously about how much I love my current keyboard. It cost nearly $200, so I better love it, but that’s not really the point. The keys feels so good under my fingers, and it makes every other keyboard pale in comparison. I really like it, but no matter what I do or how much I invest in a keyboard, nothing can change one unfortunate fact: I am a terrible typer.

I can likely think Grade 7 for this. I have rarely use computers until then, as they were not yet the mainstream item they are now. In Grade 7, my school got a huge upgrade to its computer department, and that included a weekly computer class. My teacher didn’t really know how to use a computer, and that’s putting it nicely. She was clueless, and she practically let us create the curriculum for that weekly class.

Basically, it went like this. If we finished the class assignment early, we were allowed to just play around with a computer. This was pre-Internet, but many games were still installed on these machines. Most of them were educational stuff, things like the Oregon Trail, but someone also managed to get the original Doom installed, and so everyone wanted to play that all class. First, though, we would have to finish the daily assignment.

We got to choose what we wanted to do, and I opted for the typing program. This would be something like the Mavis Beacon typing tutor, in which it tries to teach you how to type properly, using things like the home row, proper finger placement and all that jazz. Every time you complete a lesson you got a little certificate that you could print out. We had to print out a certain number of these every class in order to play games. Here’s the thing: it didn’t have to be different lessons.

I’m impressed that I still remember the name of the program I completely ignored all those years ago.

Well, technically, you’re supposed to go through a whole learning system, progressively doing harder and harder lessons. However, all the certificates look the same if you got 100% accuracy. It was something special, a gold star with your name on it. If I simply repeated the first lesson over and over again, one that simply used the ASDF keys, I could get through all the recommended lessons in less than five minutes. I printed out my certificates, handed them in and then used the rest of that time playing around. I did this the entire year, and though I got really good at playing Doom and typing with the ASDF keys, I never got good at typing.

This has continued virtually my entire life. I use about six of my fingers, and I have terrible typing technique. I generally type slow, or at least slow compared to most of my peers, or I typequickly but with about 80% accuracy. That sounds okay, that also means one out of every five letters is incorrect. That works out to virtually every word. That’s not good.

I don’t want to say my typing is a lost cause, as it has slowly improved over the years, but I was deeply impressed by a speech recognition software demo I took part in at university. This was several years ago, and it only lasted about five minutes, but it was an impressive five minutes. It not only got roughly 90% of the words right, it used the proper spelling and grammar for various words like ‘your’ versus ‘you’re’. In other words, that speech recognition software was better than 70% of high school students.

I used to edit papers. ‘Your’ versus ‘you’re’ has become epidemic.

The first speech recognition software I used was also, somewhat ironically, back in Grade 7. It was terrible, as this technology was still in its infancy. You spent so much time correcting it that it served no real purpose. The school only used it because one student broke both wrists and thus could not write or type. She got to use the speech recognition software for her essays and tests, and while was a fun thing to play around with, it was also incredibly frustrating.

The technology has really advanced over the years, especially in the last decade. Windows 7 came with a free speech recognition program, and it was pretty decent. I tried it for a little bit, but at that time I didn’t write enough to fully use it. It also didn’t seem to like my accent, which is standard Canadian English, so that didn’t help things.

As I continue to blog, I realized I was wasting so much time constantly fixing simple spelling mistakes. It also throws the rhythm off when every time you write something that little red line appears underneath it, telling you that you’ve made another silly mistake. I did a little research, and then I invested in Dragon speech recognition software, the biggest name in the field.

Wow. This thing really works.

So far, I’m impressed.

At first, there was a slight learning curve. Mostly, this involves me learning how all the commands work. I still only know a little bit, as you can use the Dragon software to do virtually everything on your computer, from opening files to the auto-filling out forms. I haven’t looked into this much, as I mostly just want to write blog posts, but it’s coming in handy.

Also, the program needed to learn about me. It analyses my voice, my intonation, my speech patterns, my accent, everything. It learns what I am saying and how I say it, and it can then more accurately reflect what I am actually saying on screen. This is a continuous, ongoing process, but after the first three or four days the accuracy really shot up. For example, it can now differentiate between what I am saying ‘chess exercise’ and a ‘chest exercise’. That took several training sessions, but now the program effortlessly does it.

That said, it’s not perfect. Not even close. I have not corrected any of my August blog posts. If you read them, you can see exactly how the program interprets my voice. You will get vast sections of perfect prose, every word exactly what I want … and then you will suddenly get a jumbled sentence that doesn’t make any sense. Actually, it’s rarely a full sentence. In general, three or four words will be mixed up, but that’s enough to completely mangle the meaning. After September, I will be going through and completely fixing all those mistakes, but for now you can look through any past blog post and see the relative accuracy of this Dragon software.

[Edit: after going through the full month, I realized many more mistakes slipped in than I thought.  Correcting them all used up perhaps all the time I saved from not typing.  That’s … much less good.]

Also, some words it simply cannot recognize certain words. I don’t really have an accent. If I do, it’s incredibly faint. I like to think I speak fairly normal. The Dragon software disagrees apparently, especially with the word ‘goal’. I had to type that, because if I say that word, it will write ‘girl’. To me, those two words sound very different. No matter how many times I say goal or how much I emphasize the oh sound, it writes girl. Strange. There are other examples as well, but I was the biggest one, as I took about 10 minutes speaking into my microphone trying to see if I could get goal to appear my screen. I could not. There is probably a joke I can write here, but let’s not go there.

Close enough.

One month of using this program has made the writing aspect fairly second nature. When I first started, though, it felt extremely weird. I was talking at my computer, and that seems alien to me. I’ve spent a lot of my life writing, and I get into a certain mental groove if you will. I constantly mutter to myself or make self-deprecating remarks as I type. I couldn’t do that here, as obviously that would then show up on my screen. I had to control this aspect of myself, and that actually made writing very hard for the first bit. I almost felt handicapped, like I was missing an important element.

Actually, the biggest adjustment was the sheer speed. Because I write or type so slow, this gives my brain a chance to race ahead. I will type out a sentence and my mind is already formulating the next one. With the speech software, that time was reduced to almost zero. I think of a sentence, I say that sentence and then that sentence appears. I had no time to think of the next one, and again, that blew me away a little bit. My initial writing samples with this were very bulky and jagged, as I was basically saying individual sentences with no flow between them.

That said, the more I use the software the faster I feel my brain actually working. It no longer needs the 15 or 20 seconds it would take for me to type out the sentence to start thinking of the next one. Rather, I say a sentence, and instantly my brain is working on the next one. It’s a cool feeling, almost like trading in your bike for a motorcycle. Everything is going faster.

And that’s really the big thing. Using this, I am writing far faster than before. I’d wager after the first 10 days, my productivity simply soared. Throughout university and high school, I average between 500 and 750 words per hour. With these blog posts, I’ve upped that to nearly 1,000 words per hour. Since using the Dragon software, I’d nearly doubled that. Of course, there are some variables here, as it’s far easier to write an informal blog post about cats than it is to write a scholarly article on Plato’s Republic, but the point remains: I am both writing and producing writing at a far greater rate.

I highly recommend this. It makes writing easy. I know several people who are excellent talkers, people who can hold a conversation with absolutely anyone, but they are terrible writers. I think people using the software can help bridge that gap. Just hold a conversation with your computer, imagine you are talking instead of writing, and I bet a lot of people can improve that way. If you are a good writer already, the improvement is simply fantastical. I love it, and I cannot even imagine going back to typing with my six fingers. That would be horrible.

Dragon software is recognized as the best, but it’s also fairly expensive. Other software on the market has less accuracy but is also cheaper. If you have Windows 7, you have a free program, though I’m not sure how good that one is. I thoroughly enjoyed using Dragon, and I would recommend it without hesitation. Just be prepared for a 10 day or so adjustment period, and then you’ll be wondering how you ever got along without it.

University students, this could really be a timesaver. Look into it.

Leave a Reply

Your email address will not be published.