22 July 2012

Cyberspace August 2012


I regularly revisit technology that I think has potential but hasn't made it yet. In the mid-90s I wasted a lot of time on voice recognition on the Macintosh, and asking the computer what time it was usually ended up in launching WordPerfect. I've done the same with Linux over the years, and a recent trial of Ubuntu 12.04 reminded me why I still don't use it.

I decided to revisit voice recognition this year, and purchased Dragon NaturallySpeaking 11.5 Premium (available for Windows and Macintosh). I normally use it wearing a headset at my computer, but there is an iPhone client which allows you to dictate as if you were speaking on the phone, and later upload dictation for transcription later. The voice recognition is very accurate, and after learning the basic set of commands I can navigate my PC, open and close applications, dictate magazine articles, and look at documents rather than the computer while dictating, just as I would do with a normal handheld dictaphone.

Is this more productive? I think it is - I recently wrote a 3,000 word paper using the system and I believe I saved approximately one third of the time I would have taken to type it myself, despite being a fast typist. I can now read, look at other documents, use both hands with a book and avoid common spelling mistakes while getting my thoughts down. It is reasonably tolerant of background noise and regularly processes the dictation it receives to create a voice and audio profile that seems to negate some of the effect of working in an open plan office. The transcription is context sensitive, so it makes fairly accurate judgements as to whether I want to use for or four, and it can insert commas and full stops for you. If it struggles with a word you can spell it and it will remember that word and all your particular pronunciation of it. The main requirement is to speak clearly, because it is not as intelligent as your secretary - however that does not mean that you need to speak like a robot. On the contrary, it prefers you to speak naturally in full sentences so that it can understand the context of the words you dictate.

Would I recommend it to another lawyer? Yes. It provides the freedom of dictation with instant results.If you are happy to go back and revise it yourself you can get a lengthy document out quite quickly. I haven't fully explored all the commands available or the customisations you can carry out, so that inserting footnotes and other rather specialised activities are still a manual process for me but I have no doubt that I will master those in time as well. I can even open an e-mail in Outlook, reply and send the reply all without touching the keyboard. It works in most applications, including browsers, so if you use Web based e-mail you can still dictate. It can also scan your e-mails to learn the names of people you correspond with, so you can say "new e-mail to Joe Bloggs" and it will understand who that person is and his e-mail address.
One particularly interesting modification you can make is to set up commands that will automatically insert boilerplate content or signatures and logos. For example, you may have a standard limitation of liability clause for trusts – you can insert a whole page with one command.

Both Windows 7 and the next version of OS X, Mountain Lion, have built-in dictation abilities. However, they are not sufficiently advanced to use in practice. You may be able to make them work, but the cost in time will far outweigh the cost of buying a dedicated product.