Wozniak is 69. You can do your own grim maths calculation here. A self-driving car may yet pull up in my lifetime, hopefully your’s too.
The tech sector has a long history of misplaced ‘coming real soon now claims’.
One of my first jobs covering technology was in 1981. I went to a press conference showing an early speech recognition computer. It could just about understand ten words some of the time if you spoke very carefully.
At the press conference we were told computers able to recognise and understand everyday speech are just two years away. They’ve been just two years away ever since.
Self-driving cars are not that different. In fact the reason for misplaced optimism is much the same. That is, people are terrible at forecasting future technology.
Waymo, which is part of Alphabet (Google) has been testing driverless taxies in Phoenix Arizona this year. Waymo choose Phoenix because it has wide, flat roads.
In theory it is one of the easiest places in the world to drive. In practice Google still sits human drivers behind the wheel; just in case.
One reason for overconfident forecasts is that tech company leaders believe their own hype about progress in artificial intelligence and related technologies.
Progress is difficult. Much of today’s AI uses brute force; improvement can be a long, hard slog. That doesn’t sound anything like as good at a rah rah sales event as whipping up excitement about what could be possible.
It needs to deliver: an annual subscription costs a NZ$240.
At that price Dragon Anywhere is not a buy, try, forget app store experiment. It’s a significant investment. It needs to earn its keep.
Worth the money?
For some people Dragon Anywhere will be worth every penny. Accurate speech to text software can unpack a new level of productivity for some people. Not everyone will see a return on the investment.
If you already use desktop dictation software, you’ll have an idea of what Dragon Anywhere can do for you.
Being able to dictate text to an iPhone is a bigger deal than it might sound at first hearing.
The designers made the iPhone for dictation. Let’s face it, writing on a tiny glass keyboard is a challenge if you want to do anything more than send a text or a tweet.
I’ve written 1000 word stories on the iPhone. It’s not fun, nor is it productive. The alternative to dictation is carrying a Bluetooth keyboard. That can be a pain in the backside.
It also means you can replace desktop dictation with your iPhone. Given that your phone goes everywhere you do, it means you can produce text almost anywhere. This explains the product name.
You could, for example, write while in the back of a car or lounging in bed. In practice I found using the iPhone for dictation is more natural than using a desktop or laptop Mac.
Mobility is important, because ideas do not work nine-to-five in an office. Your writing muse can turn up unannounced at any time. With Dragon Anywhere you can jot down your ideas as they appear. There’s no need to hunt around for a computer or a pen and paper.
Your phone is already your most important computer. Dragon Anywhere takes that further. Depending on how you work, you may be able to ditch the desktop altogether. Although if you don’t want to, Anywhere integrates with Nuance’s desktop dictation applications.
If Dragon Anywhere save you buying a new computer, the subscription starts to look like a bargain. Even if you don’t go that far, your typewriter keyboard may gather dust.
Dragon Anywhere works where there’s a connection
The software doesn’t quite work anywhere. You need a live internet connection. Dragon Anywhere calls on Nuance’s cloud resourced to work its magic. That means you can only use it when you have a live internet connection.
The good news is that it sips data. You might run through a megabyte or so dictating thousands of words. I found after an hour’s use, my data consumption was still measured in hundreds of kilobytes.
Another piece of good news is the cloud round trip is fast. Speak a sentence or two, pause and the text is there on screen. It takes seconds. I found I couldn’t dictate fast enough to get ahead of the cloud connection.
In other words, you can use Dragon Anywhere while you’re on the move. If you have anything but a minimal data plan you can use it without counting the bytes or hunting for Wi-Fi.
Nuance says it encryopts connections, so criminals can’t listen in on your dictations.
How well does Dragon Anywhere perform?
The performance is impressive. I used it to write a first draft of this review. From the first words I uttered it was catching almost everything without error.
The software stumbled over the word iOS in the first sentence. To be fair, it’s a specialist word. If you think of how you say the name: eye-oh-ess, not picking it up it understandable.
It wasn’t the software that stumbled in the second paragraph. I can take the blame for not figuring out how to say NZ$240 in a way that made my meaning clear. Put this down to user error.
The third sentence was perfect.
Out of the first hundred words, Dragon Anywhere got everything except iOS right. That’s impressive. Remember this was my first try of the software. The software had not encountered my voice or accent before.
In practice it learns a little as it goes along. To see how this worked I read the words again and this time Dragon Anywhere scored a perfect 100 percent. It understood iOS. The software understood my speech far better than Apple’s own Siri software.
If you make an error, fixing your text is easy. The only barrier is that you have to memorise instructions. In most cases the words are obvious, you don’t need to guess them. Some take a little practice.
I ran into a problem with some New Zealand place names. That’s understandable. Dragon Anywhere allows you to add custom words to the system which gets around the problem.
The productivity question
If you notice, I hedged my words when I said the software could be worth the money. Likewise when I said it may transform how you work or make you more productive.
That’s because, good as it is, speech recognition is not for everyone. In my experience it takes longer to dictate stories than to type them. I also find I struggle to compose while speaking. This could be down to 40 years of touch typing. With practice my dictation speed might improve.
There are also times where I need to write and dictation isn’t the best tool. Writing on a train, an airplane or somewhere public would be too much for everyone else.
If you find typing is difficult or run into overuse problems, then its a godsend. If you think by speaking, you’ll love it.
You have to hand it to Nuance. The latest Dragon Professional voice recognition software is impressive. So is its business model.
Dragon’s software is expensive by today’s standards. A single user licence for Dragon Professional version 6 costs US$300. You can buy a PC for less.
Keep in mind both Apple and Microsoft include voice recognition software as part of their operating systems. Neither application is free, but they are already paid for. So, in effect, Dragon is, asking people to pay again for something they already have.
That puts the business’s livelihood on the line. To make it work, Dragon has to offer something special. It does that. Dragon Professional version 6 performs far better than the alternatives from Apple and Microsoft. It uses something called deep learning to improve accuracy.
Dragon Professional more accurate than alternatives
Dragon claims this means the software has 99 percent recognition accuracy. There’s no easy way to verify the claim, but in testing the software does a near perfect job of turning spoken words into computer text.
What the numbers don’t tell you is that even a small improvement in voice recognition accuracy means a vast improvement in the experience. The difference between going back and correcting every tenth word, 90 percent accuracy, and every hundredth word, 99 percent accuracy is huge.
You can improve performance by training the software. If you’re committed to using Dragon, then investing some time makes sense. Yet in practice the software works so well out of the box you might decide to just get on with it.
It seems the software does some form of training when in use. Every so often there’s a message to say Dragon is updating your profile.
Dragon Professional isn’t a stand-alone application. It is integrated into the operating system. It works with other apps. Apple Pages and Microsoft Word are the obvious candidates, but any program using text input should work.
It works with almost every Mac app that uses keyboard input. In theory you can control less word-oriented apps such as image editors, but that doesn’t make practical sense. Having said that, Dragon Professional is excellent at performance everyday MacOS commands. It would be an ideal tool if you had access problems with your hands.
A small icon appears in the Mac’s menu bar, in much the same way as other system level apps. When the software is in use a small floating window opens on the screen with three more icons. There’s also a guidance window with help when you need it.
For most of the time the second window keeps out of the way. The main one is small enough to not be a distraction. The microphone icon shows green when it is on and red when it off, otherwise there’s not much to see.
If you like using the cursor you can switch between Dictation, Command, Spelling and Numbers mode using the main floating window.
There’s a transcription mode which allows you to turn audio files into text. It’s a lot more hit and miss than the normal dictation software. It manages to cope with a few minutes of audio where there is only one speaker, but chokes if you attempt to transcribe, say, an interview with two people talking.
A personal productivity note
While the technology in Dragon’s latest voice recognition software is impressive, it’s not for me. After forty years of touch typing, I write with my fingertips. Any attempt to compose faultless prose using my voice ends in an embarrassing mess.
That’s not to say I don’t make typing errors. Anyone who has read my tweets can see that. Yet the flow of my writing is so much better when I hit keys than when I speak. No doubt that would change if I spent thousands of hours improving my technique. Simple economics says I’m better off sticking with what I know.
Take this blog post as Illustrating how this works. I started out trying to use the Dragon Professional software to write the post.
It was a disaster. Although professional scribes are taught to ‘write like you talk’, that advice is not meant to be taken literally. When I gave up on voice recognition, I hit the ground running and finished the post in minutes.
There’s another problem that may affect some readers. I feel self-conscious and uncomfortable when dictating to a machine. There’s nothing worse than knowing people can hear you as you compose a story. That’s not an issue when I type.
While we are on the personal stuff I should mention another major plus for Dragon. Most voice recognition tools struggle with my accent. It’s a hybrid British-antipodean thing.
UK voice recognition settings don’t work for me. Nor do New Zealand ones. Oddly, Australian settings get me the best result on Apple equipment. Nothing seems to cope with my voice on Microsoft systems. Dragon Professional worked out of the box even though the settings are hard-wired to New Zealand.
All the above leaves me in an odd position. I’m about to recommend a product that I wouldn’t normally use myself. So let’s run through the main points again. Dragon Professional version 6 does an excellent job of turning spoken words into text. The software is accurate and reliable. It also provides a great way to control a Mac when you can’t or don’t want to use hands.
Apple may have something special, er, up its sleeve to change this should it decide to launch a smart watch.
In the meantime, one thing could push smart watches into the mainstream: Voice recognition.
Smartwatch makers need something like Apples Siri, only more so. Or, if you prefer, like a swept-up version of the speech commands used with Google Glass.
Smartwatch makers need to use voice
Instead of squinting at a tiny screen for information, a next generation wearable device can speak the information. Today’s speech technology is good enough for this to sound almost natural.
Decent voice recognition means no more struggling with swipe gestures on a tiny display. No more ridiculously trying to type on a minuscule onscreen keyboard.
Now here’s the rub. When the screen becomes less essential, the device doesn’t have to sit on your wrist. It could become a badge — like the communicators Star Fleet officers use in the Star Trek TV shows.
You could ask questions: What is the time of the next meeting, when does the next train arrive, what is the square root of 37?
Speaking to a huge audience while live on the radio was an important step to overcoming my fear of public speaking.
It is one thing to speak to a microphone and one or two others in a studio. Standing in front of a crowd or on a stage is something else.
I cracked it by starting small with:
a small crowd,
a short talk
an informal setting
and an easy topic.
My first official public speaking engagement was for a local computer company in Wellington, New Zealand. The company held regular evening events with food and drink for its customers. It would invite speakers and get them to talk informally about a technology subject likely to interest the audience.
Find a public speaking subject you know about
The company’s managing director asked me to talk for about 20 minutes on my work as a technology journalist. I can talk on this subject about for hours.
Although I prepared and wrote out a list of topics to cover, I didn’t have formal notes. It turns out these are more of a hinderance than a help. and some nervous people are too ready to hide behind them.
Today some hide behind Powerpoint slides. This can make for an awful audience experience.
Thankfully, the audience was unintimidating, maybe 25 or so people and the mood was friendly.
When I finished a discussion started which went on for almost another hour. I didn’t realise until afterwards that I chaired the discussion keeping things moving along when it flagged. I just did what came naturally.
Over the next few months I had a few similar speaking engagements, the audiences remained small, but the session length and topics would be different. After around five or six events I was comfortable enough to tackle a larger crowd.