Voice-recognition software has Hugh John wondering about the future of his keyboard
Nuance's innovative technology and astute dealings in the marketplace have established it as the principal speech recognition software provider. A licensing agreement with speech pioneer IBM (Via Voice, anybody?) and the recent acquisition of MacSpeech, means that the company's Dragon Naturally Speaking will become the software of choice for most Windows and Macintosh users.
Version 11, released recently, should consolidate that position. Nuance claims that the new Dragon, whose engine has been redesigned to take advantage of developments in processing technology – principally multi-core tasking and 64 bit – is faster, smarter and more accurate. And the Dragon is spreading its wings. Not only is it available to Windows and Mac computer users, from netbook to server, but iPad and iPod Touch owners have taken enthusiastically to the new, free Dragon apps.
Dragon Dictation and Dragon Search apps (more here) now enable Apple and Symbian users of mobile devices to quickly and easily speak any text, whether it’s for "a status update on Facebook or Twitter, to send text and email messages, quickly look up a recipe or the latest YouTube video, and everything in between". Both apps are available on the UK Apple App Store free for a limited time.
One of the most impressive features in the new version of Dragon is the enhancement of the Sidebar, now resizable and dockable. Although Nuance had previously made big improvements in simplifying the program's ancillary voice commands – 'insert', 'save as', 'check grammar' etc – the sheer number of commands involved required either constant use or an excellent memory. The Sidebar now squats by default on the right of the screen, context sensitive, and ready to help with commands and tips. The four tabs – Home, Global, Mouse, My Command – have drop-down expandable menus which contain all the voice commands likely to be used. It is also possible to create personalised voice commands for repeated
Calling up the Dragon Sidebar makes dictation considerably easier and enables quick and confident navigation through menus. The sidebar has commands for Microsoft Word, WordPerfect and Microsoft Office including office Word 2010. Readers who rely entirely on voice commands should find Dragon 11 a big improvement.
As with previous versions, the recognition software takes time to adjust to one's particular speech patterns and articulation. In version 11, however, the training time has been drastically reduced. Version 11, says Nuance, 'learns' better than its predecessors with, "new recognition analytics that constantly monitor audio quality". In fact, after frequent use, it becomes hard not to think of Dragon as some sort of ersatz desktop friend. On more than a few occasions I've found myself chastising the beast for its rare recalcitrant behaviour: "Dragon, how many times? There's no such thing as 'Stir Friday’; I said, 'stir fry'".
Any mistakes, it should be pointed out, can be rectified in a number of ways and Nuance claims that each time a recognition error is addressed the user reduces the chance of it being made again. Once the user becomes used to talking and is more relaxed, results improve. There's no need, for example, to let the on-screen text 'catch up' with your dictation and Nuance advises that speech should be natural and, where possible, flowing, as speech recognition software in general is contextual. The longer the phrase or sentence the more accurate the speech recognition.
Transcribing from digital voice recorders couldn't be easier
Users can choose the most comfortable and convenient way to edit transcribed text. Some may want to talk without interruption and do their editing at the end of a session while others may want to use the mouse and keyboard to edit while dictating. Either method is available and Nuance suggests that new users take some time to experiment with when and how to correct errors.
Transcribing from digital voice recorders couldn't be easier, with the minimum voice training enrolment having been reduced from 15 to five minutes. Speech files can be imported into a computer via cable or memory card. Simply
drag the sound file icon on to the DragonBar and, without even being prompted, Dragon will do the business. 'Transcribing, please wait' appears on the screen and, bingo... one sound file rendered into a text file. Now that really is impressive and it's a feature that should prove very useful for students and teachers.
Incidentally, Dragon 11 can now accept .dss and .ds2 formats (found on recorders from Olympus, Grundig and Philips) directly. Files recorded using these digital speech recording formats previously required conversion prior to transcription. It can also handle WAV. and MP3 formats.
We're not yet at the stage - well, this reviewer isn't - where dictation has replaced the mouse and keyboard. But Dragon 11 represents a discernible shift in the communication paradigm, and with Nuance, Microsoft and Google all ferociously driving the pace of speech recognition development, it may not be long before the mouth is mightier than the fingers.
Ratings (out of 5)
Fitness for purpose 4
Ease of use 4
Value for money 4
Dragon Naturally Speaking 11
Speech recognition software for Windows computers.
Prices: home version, £70; premium, £120; Bluetooth version, £200; Premium with Philips digital recorder, £200
Available direct from Nuance or from Amazon UK.
(Mac products powered by Dragon – Macspeech Dictate £180, MacSpeech Scribe £120 available direct from Nuance.)
Nuance Education Page
Nuance's See and Speak Accessibility Suite
Useful matrix supplied by Nuance for assessing performance of third party headsets, microphones and voice recorders with Dragon Naturally Speaking