Jack Schofield 

Can speech recognition software help prevent RSI?

Sean’s wife spends up to eight hours a day typing reports and he wonders which software works best. But speech recognition is already built into Windows
  
  

Bandaged hands on keyboard
There are several causes for RSI, including badly designed devices and bad working practices. Photograph: Zigy Kaluzny/Getty Images

My wife can spend up to eight hours a day typing reports in Microsoft Word 2010 and due to RSI issues, is thinking about investing in some speech recognition software. There seems to be a huge diversity in the price of these, and we need to know how much we should spend and how reliable they are. She would also be interested in a portable version, which she could use on the go.

I’ve been on a few websites and I’ve identified the Dragon Naturally Speaking 13 Premium Wireless as a possible solution. We’re running Windows 7 with the option to upgrade to Windows 10. Sean

There are at least four issues here: technology, usability, ergonomics and economics. The technology aspect is pretty simple: if you want to buy a speech recognition program, Dragon Naturally Speaking has been the most accurate system since it first appeared in 1997. However, it doesn’t matter how accurate it is if your wife ends up not using it.

Fortunately, you already own a good speech recognition program. It’s built into Windows 7 and other current versions of Windows. It’s part of the accessibility suite through which Microsoft caters for people with various disabilities. Other components include a text-to-speech program or screen reader and a magnifier.

Your wife should use the built-in program for a week or three to see if she can incorporate it into her workflow. If it does the job, that solves the problem. If it works well but she would benefit from more accurate transcription, then it would be worth investing in one of the versions of Dragon Naturally Speaking, which cover a wide price range. But if it turns out that she can’t benefit from Windows Speech Recognition, buying a better program would be a waste of money.

Using speech recognition

To run the Windows Speech Recognition program, type the word speech into Windows 7’s search/run box: it should be the top result. It won’t run if it can’t find an audio input device, but almost every laptop and tablet has a built-in microphone. If it does run, a wizard will take you through the setup process, which involves reading a short text aloud to check the microphone.

Once set up, Windows Speech Recognition runs in the background, but you can control it with either voice commands or mouse clicks. It’s designed to be an assistive technology, so it’s easy to control via spoken commands such as “start listening”, “open speech dictionary”, “comma”, “period”, “correct”, “delete”, “save” and so on. SchoolFreeware has a series of YouTube tutorials, and the fifth one provides a good guide to dictating text into Word.

However, Windows Speech Recognition will work with any program, including WordPad, the free word processor included with Windows.

Training matters

Speech recognition has improved dramatically thanks to the use of statistical models. Programs that struggle to recognise isolated words and letters can use a vast repository of stored language to get whole sentences correct. This is very evident with the simple stuff that Microsoft Cortana, Google Now and Apple’s Siri almost invariably get right nowadays. Longform dictation is somewhat trickier.

Like its rivals, Windows Speech Recognition gets better the more you use it, because it learns your vocabulary. SchoolFreeware reckons you can get to around 140 words per minute, which is faster than most people can type. However, as with touch-typing, speeds slump dramatically if you have to make corrections.

People who don’t speak slowly and clearly, and who don’t put in the time to train the software, usually get discouraged and give up. The people who use speech recognition successfully are usually highly motivated, often because they can’t use their hands at all. This may be because of their job situation (eg dentists and pathologists) or because of RSI.

Enter the Dragon

In my experience, Windows Speech Recognition works very well when you read out proper texts. It’s not as good if you’re making stuff up as you go along, and frequently need to rewrite sentences, or if you use a lot of jargon. More advanced programs such as Dragon Naturally Speaking 13 can save and reload user profiles, and can import vocabularies for specialised fields such as medicine and the law.

Dragon Naturally Speaking can also convert dictation files. That means you can dictate to a handheld digital recorder or other source and then load the audio file for processing. In fact, the Premium Mobile version includes a small voice recorder for people who want to dictate texts while on the move. Of course, you can use other recorders but you need to create a profile for each one.

There are cheaper versions of Dragon Naturally Speaking: for example, you can get excellent results with Version 12 Home for £34.55. But for business users, the price is irrelevant. The £129.99 Premium 13 Wireless version, which includes a Bluetooth headset, can pay for itself in a few days.

As usual, there are alternative programs but frankly, you may as well use the free one in Windows or pay for Dragon.

One that might be worth a look is voice typing for Google Docs: you can dictate into an online document using a microphone and the Chrome browser, then download the document and load it into Microsoft Word. While on the go, you can dictate texts using the Google Docs app on a smartphone, but when I tried it on my Android phone, it was a nightmare. It got words wrong and it didn’t recognise commands such as space, full stop, period, or delete, so my text included sequences like “full stop delete full stop period correct period”. Also, the mic kept turning itself off.

Ergonomics matter

RSI (repetitive strain injuries), RMD (repetitive motion disorders) and carpal tunnel syndrome don’t come from nowhere. They are caused by badly designed devices – such as laptop computers – and bad working practices, such as not taking regular screen breaks.

I’ve written about this several times because I’ve needed two sets of expert physiotherapy to survive, and other people have needed surgery. The problem is that everyone thinks they’ll be fine until they aren’t, at which point it’s too late.

It’s important to use an ergonomic split keyboard for sustained work (at this instant, my B key is 7cm from my N key). To avoid “the vulture posture”, it’s important to have the screen raised to the correct height, which means using an external keyboard with a laptop. It’s useful to remember keyboard commands – they can do almost everything in Windows – to avoid using a mouse. (I’m currently trying a Penclic, which is still a mouse but pen-shaped, but a digitising pen would be another option.) It’s important to take regular screen breaks: I’m now using a Microsoft Band 2 to buzz me reminders.

For more details, see my earlier answer: How can I use laptops and tablets without suffering from physical pains?

RSI can be a temporary inconvenience, but depending on which type it is, it can also be a life-changing and debilitating problem. Working through pain is the wrong answer. Speech recognition, improved ergonomics, screen breaks, gentle exercises (Tai chi, yoga) and physiotherapy can help, but if she’s in pain, your wife should see her doctor.

Have you got another question for Jack? Email it to Ask.Jack@theguardian.com

 

Leave a Comment

Required fields are marked *

*

*