Nuance has recently released the latest version of their popular voice recognition software, Dragon NaturallySpeaking. Nuance claim up to 99% accuracy with this latest version, and the ability to produce text at up to 120 words per minute (3 times the recognised average of 40 words per minute for competent typists). Combining this with the ability to control the mouse by voice, carry out Internet research through simple voice commands and working quickly and effectively across a range of Microsoft programmes, it sounds amazing – is it?
Well, there is no such thing as a perfect solution, no matter what technology is being looked at, but for those of you who last looked at Dragon NaturallySpeaking at version 7 or older it is definitely time to look again.
It will take a confident and experienced PC user, confident in what they want to say, to be able to achieve the kind of results Nuance are claiming. Nonetheless, Dragon NaturallySpeaking can still do a job for the less experienced user.
Dragon NaturallySpeaking has long been viewed as an effective tool for those who find that using a keyboard can be difficult – there is no doubt that v10 remains an effective tool in this context. There is a range of opinion as to its use by those who have literacy difficulties – reading or producing text. The argument goes that such users would not be able to easily pick up when Dragon NaturallySpeaking makes a mistake, and that the spellchecker cannot pick up these mistakes as they are not spelling errors. There is truth in this, but v10 has developed the text-to-speech functionality to help counter this. By selecting the text and saying ‘play that back’ the user is able to listen to what it is they said. By then saying ‘read that,’ Dragon NaturallySpeaking’s voice engine will turn the text into speech, using ‘British English Jane’ who will be familiar to textHELP users as a voice that is a big step on from the Microsoft Mary’s of this world. The user can then listen to the actual text that has been produced, and compare it to see if that is what they actually said. It can then be quite straightforward to ‘correct that’ and effectively produce work using language and terminology that may not otherwise be available due to spelling or structuring issues. In addition, ‘Jane’ can also be deployed to read any electronic text that is available in a Word document – similar to one of the best tools in textHelp, enabling increased accessibility for reading tasks.
No panacea, and sure it will take time to listen back to the text produced, but if the text production is three times faster than typing, and the accuracy over 90%, then for some users with literacy difficulties this could prove to be an effective tool in reducing some of their barriers to producing written work.
Perhaps it is worth reacquainting yourself with Dragon NaturallySpeaking.