This reply is being dictated using Dragon NaturallySpeaking version 7, Professional Solutions (I have also used the cheaper Preferred Solutions version of NaturallySpeaking version 7).
IBM ViaVoice is now being supported by ScanSoft, which bought Dragon NaturallySpeaking from Lernout & Hauspie. You might want to check whether ViaVoice will continue to be developed after the present release as a continuous speech recognition product. I cannot comment on continuous speech recognition products for the Mac but you will find some advice in the forum at the first Web address below.
In my experience, NaturallySpeaking version 7 is a very effective tool for producing correspondence, reports (including scholarly papers), and e-mails. It works well with most Microsoft products and WordPerfect. There is an active and supportive user community for NaturallySpeaking which you can access via http://support.lhsl.com/databases/dragon/sitereg.nsf.
Speech to text software such as Dragon NaturallySpeaking is wonderful for persons who have disabilities or difficulty keying large volumes of text for reasons of repetitive strain injury. It does, however, require a reasonable understanding of how software, hardware, and operating systems work together, and well specified kit. For example, do not even try to run Dragon NaturallySpeaking 7 if you have less than 256 megabytes of RAM (512 megabytes is better); you need a Pentium III 600 megahertz processor or better; a fast front-end bus; a fast and large hard drive; a good-quality soundcard; and a good-quality microphone. You also need to learn how to dictate consistently. For further advice you could try the forum above and these addresses http://www.emicrophones.com/ in the UK, try http://www.itspeaking.com/ I have no commercial or other connection with either of these companies but the proprietors of each have been generous in their time and advice when I have asked for support].
Using standard paragraphs for comparison purposes, many users report they can achieve better than 99.5 percent accuracy. I can certainly get better than 99 percent consistently (ie less than one word misrecognition per hundred words dictated).
All speech recognition products rely on recording an extended sample of speech, analysing it, and using it for comparison purposes with dictated text to match sound to text. This process is usually referred to as "training". Speech recognition products also use context recognition to match speech utterances and discriminate between homophones (deer, dear, etc). Speech recognition software works well for single users who have invested time in training the programme. Such software is, however, not good at discriminating between and accurately recognising the speech of several different voices, so you cannot simply feed in recordings from a focus group session, for example, and expect the software to transcribe it.
I have used previous versions of NaturallySpeaking to edit and control QSR N5/N6 but I cannot comment on how well it works with other qualitative research software.
Hope this is helpful
Dr David Cairns, Assistant Director, Reviews Group, QAA
Southgate House, Southgate Street, Gloucester, GL1 1UB
Tel: 01452 55 7106 (direct line) Fax: 01452 55 7011
Mobile: 07774 816875
PA Mrs Tracey Hendry 01452 55 7108 e-mail [log in to unmask]
From: Christian Nelson [mailto:[log in to unmask]]
Sent: 05 September 2003 16:06
To: [log in to unmask]
Subject: comparison of DNS 7 with Via Voice
Is anyone able to make a comparison btwn Dragon Naturally Speaking 7 and Via
Voice for Mac (latest version) in terms of quality/efficiency? I know DNS
doesn't run on Mac, but I have access to both Macs and PCs.
Thanks for any help,
If you receive this message in error, please inform the sender and delete the message. All contents have been scanned by QAA, although this is not a guarantee that they are virus free.