I keep thinking about Speech Recognition being "a mystery" and a "surprise" to so many people when they hear about it. It's as if I am talking to someone about driving a car, and finding that they think you can only get around by riding a horse. Or using candles instead of a light bulb.
Okay. The car has been around for quite a few years, and most people know you don't have to ride a horse or have one pull a wagon you ride in. Likewise, the light bulb. But speech recognition (or "voice recognition") has also been around and working well for more than 10 years. And I'm surprised again and again at how few people use it.
Maybe it's because no one knows how well it works. And the original bad impression we all had in 1999.
As of 2015, you can easily say 80 words per minute, talking into a good microphone headset. Talking into ANY of the programs that you use all day: your MS Word, Excel, Powerpoint, Email, Facebook, and other programs.
And while you talk into your headset, your mouse and keyboard still work, if you feel like using them. Even though you find that, like most people, you would like to just think and say your thoughts out loud and see them land on the page, instead of doing a lot keyboard work, wearing out your fingers, and making the usual typing and spelling mistakes.
Guess what? 80 words per minute, at least, is how good speech recognition is, if you get any version of Dragon NaturallySpeaking since 2003. You can "stream" the typing right out of your thoughts--what we usually call "talking." You throw paragraphs onto the page at 80 words per minute or faster. With amazing accuracy. But so many people are unaware of this, and typing and mousing and working hard to put their words on the page, instead of easily doing it with a microphone headset.
Here's a perspective on typing speeds:
A two-finger typist can get a speed of 10 to 30 words per minute, if they work hard. An entry level clerical job (or the typing test at a job agency) requires a speed of 30 words per minute. A higher-level document worker, such as a legal secretary, must have a speed of 60 words per minute or higher. And the highest document workers--the people who complete legal documents all day in a law firm word processing job -- do 80 to 100 words per minute: 10 single-spaced pages per hour, or more.
You can do 80 to 100 words per minute by talking into your computer with a microphone headset, and a good program like Dragon NaturallySpeaking.
So I wonder, and you might wonder --
If we can all "type" 80 words per minute just by talking, why is anyone not using speech recognition? Especially students who have to get all those reports written. Especially business people who have to get a lot of proposals and communications out to their clients. Especially lawyers, paralegals, secretaries and law school students? Hey, how about teachers who have to get out all those lesson plans, assignments, and other paperwork?
How about people who just want to write a few emails or Facebook comments, quickly?
Here's why speech recognition -- and it's #1 program, Dragon NaturallySpeaking -- isn't being used by all those people.
In the late 1990's, the leading word processing program** included a speech recognition program for free when you bought a copy of WordPerfect. And people throughout the document world -- especially the law firm world -- were excited! Everyone bought the new WordPerfect and got ready to conquer their documents with Dragon NaturallySpeaking. Yay! We can just talk -- we can dictate long documents right into the computer! Fast! Painless!
I was one of those people excited about this. I was being paid by law firms, to write summaries of depositions for attorneys. If I had a 300 page deposition, I was supposed to "sum it up" in about 30 pages. This was actually very intense work: You read and you write and you never took your eyes (or your attention) away from the 300 pages for a few hours. You read while you keep your hands typing; and if you take a break, the money you earned that hour might go down from $30 an hour to, say, $10 or $5 an hour, depending on the break. Likewise, you better not answer a phone call or let your attention drift onto something else. It was extremely demanding work. But hey--here comes the easy times: speech recognition! You can sit back, relaxed, and read through the 300 pages and just talk your microphone headset right into the computer! Wow! Easy! No more backache! No more handache!
Yes, I wanted -- and every working writer, lawyer, teacher, and student wanted -- to use Dragon NaturallySpeaking. Good times are here.
Minor problem: It didn't work.
As of 1999, Dragon was terrible. Every three or four words, it misheard you. It kept mishearing you, and you kept trying to say things better. You spent so much time saying things over and over again -- A -GAIN --- AAAA--GGGGAIIINNN ... AAAAAAAA----GGAAAAAIIIIIINNNNNNNNN -- oh ^#&@(*^%&%@)(!%+$&*%%&* *&*&&^%&^% ----
And you'd look up and see that instead of again, it typed "egg hen."
Like everyone else, I stopped using the program. It was just a time waster.
And like everyone else, I didn't want to hear about Dragon again. And most of us didn't know that by 2003, a different company had bought Dragon, and fixed it. They made it amazingly accurate.
For starters, it heard "again" instead of "egg hen." And, you didn't--have--to--say--each--word--very--slowly--any--more. More. Morrrrrrrrrrrrre....
You just talked like a normal person, and it worked like it should, and it was wonderful. As of 2003.
And nobody knew it.
Nobody wanted to hear about Dragon again. We'd all been disappointed, and we all had a pretty sour response when it was mentioned.
Fact is, as of 2003, Dragon NaturallySpeaking could hear most of what you say the way people hear you. Just talking in the normal flow of a conversation, instead of slowly saying each word.
Unlike the original word--by--painful--word edition--
If you talk to someone and you say "the boat traveled down the Mississippi," they're probably going to know that you said Mississippi. They know you didn't say "this is sipping" or something else ridiculous. But in the big exciting 1999 release, Dragon often heard something ridiculous like "this is sipping." When the new owners fixed it up, it understood you in the context of all your words; and now it "realized" you had said Mississippi, so that's what it typed. You and I understand each other's words in context, and as of 2003, that's how Dragon NaturallySpeaking understood and typed what you said. So the mistaken words and spellings pretty much went away.
If you said "we'd like a table for two," nobody would think you said "weed like a table four, too" -- and neither did Dragon. People know what you said; so did Dragon. As of 2003, Dragon would almost always know what you said and type it for you.
But in 2003, most of us didn't know the program was fixed. So we kept typing everything. We kept shying away from Dragon and other speech recognition programs. And that's why most of the world is still not using this program that was good and 2003, and amazingly good in 2015.
Just like everything else in the computer world, Dragon and other speech recognition programs have improved their accuracy and speed every year, whether or not people are aware of it. Hey, just think about how good Siri is, on your cellphone.
As of 2015, if you use Dragon NaturallySpeaking, you can "type" 100 words per minute or faster. With 95% accuracy. At worst, you get 80 words per minute if you have a typical Arizona or California "accent" (no accent). And gee, what if you only had a "slow" speed like the 60 words per minute required of a secretary, or the 30 WPM required of a clerical worker?
And you aren't using this?
** In 1999, WordPerfect was still the leading program for word processing with serious document enterprises such as law firms. Microsoft Word was in second place because of its free inclusion with new computers. Soon after, WordPerfect's superior document capabilities was abandoned by most law firms and most of the world in favor of the less powerful but everywhere-advertised Microsoft Word.
Joe Loesch teaches computer skills in private and group lessons for LRS Computer Techniques.