Price:
 (Suggested retail price is $199.99. Price Watch has it for $149.  A standard version is available for $99.99, with a Price Watch price of $83.83.)

System Requirements:

• Intel® Pentium® / 1 GHz processor (for example, Pentium® M, Pentium® 4), or equivalent AMD® processor - Faster processors will yield faster performance
• 512 MB RAM or 1 GB RAM for Windows Vista™
• 1 GB free hard disk space
• Microsoft® Windows® XP (SP1 or higher) Home and Professional, 2000 (SP4 or higher), Vista
• Creative® Labs Sound Blaster® 16 or equivalent sound card supporting 16-bit recording
• Microsoft® Internet Explorer 5 or higher (free download available at www.microsoft.com)
• CD-ROM drive (required for installation)
• Nuance-approved noise-canceling headset microphone (included)
• Speakers (required for playback of recorded speech and text to speech features)
• A web connection is required for activation
• For Bluetooth wireless microphone support, please visit http://support.nuance.com/ compatibility/

During the install process the software checks to make sure your system meets these minimum requirements. If you do not meet the requirements, the software will not be installed.

Nuance Communications Inc. , Burlington , MA .  The manufacturer’s web site is www.nuance.com.  The website for Dragon Naturally Speaking is www.nuance.com/ naturallyspeaking/home/.    

Support: Except for one free phone incident, technical support by phone or email is on a fee per incident basis. There is an online knowledge base and user forum.

Dragon Naturally Speaking 9 Preferred

by Larry David, Thousand Oaks PC Club  

Dragon Naturally Speaking 9 Preferred software is voice recognition software that allows the user to talk to the computer using a normal speaking voice.  The software includes the ability for the software to learn the nuances of each user’s speaking style (referred to as training the software).  Versions are available for different styles of English as well as other languages.  The version tested was version 9.51.001.009, which implies at least 8 prior versions and suggests the product is fairly mature.   The website states: “Dragon Naturally Speaking Preferred 9 gives small business users and PC enthusiasts the power to create documents, reports, e-mails and more—all by speaking! Fast, easy and amazingly accurate, it’s over three times faster than typing. Just use your voice to dictate and edit in Microsoft® Word and Excel®, Corel® WordPerfect®, and virtually all Windows®-based applications. You can even dictate into a handheld device when you’re away from your PC, or use a Bluetooth microphone! It’s never been easier to use—no script reading required, so you can get started right away!”  The DNS software requires at least the following (or equivalent):

Intel Pentium 1 GHz processor, 512 MB or RAM, 1GB free hard disk space, Microsoft  Windows XP (SP1 or higher), Creative Labs Sound Blaster, Microsoft Internet Explorer 5 or higher, CD-ROM drive (for installation). Speakers, Nuance approved noise canceling headset microphone (included), A Web connectio n

{This is a product review of the voice recognition software, Dragon Naturally Speaking 9 Preferred.  This review is created by dictating into the computer using Dragon software.  I will use square braces to indicate what Dragon Naturally Speaking 9 Preferred heard erroneously and displayed. (The text to indicate what I said will precede the square  braces.)  Here in after, Dragon Naturally Speaking 9 Preferred with is referenced as DNS for brevity.  I will also use parentheses to indicate a side comment.  ). (In order to show where the dictation began and ended I added a bold curly brace ({) at the beginning and the end.)  

The review was begun by reading the box, [in the Pope] (never figured out what it might have heard that caused this to be displayed) inventorying the materials, and proceeding to the installation of the software.  Installation was smooth [smoother] and predictable, with no surprises at all.  It took about 20 minutes to do the installation, which is reasonable considering the software resided on two CDs.  

The first problem I observed was that the headset [and] was too small for my head.  Having a headset on for any length of time causes of pressure and pain on the side away from the headset where the pad is located.  The head band broke the third time I put it on my head.   (This may be by design, to encourage people to buy the optional Bluetooth wireless headset.)  

One step explained in the Quick Start guide involved activating the software.  This was simply a matter of being connected to the Internet and completing the form that presented itself.  According to the Quick Start guide, if you do not activate the software after five accesses of this software, the program will turn itself off.  (The software was not registered.)  

There was a short training session, where you can choose which text you want to use for training the 80s voice recognition software.  I chose the [to] Dave Barry text, thinking it would be somewhat humorous [,] and it was.  The software appeared to be able handle my chuckling with no problems.  That training took about 10 minutes.  The next step was [with] what it called writing styles [,] and it warns you that this would take from five to 20 minutes.  In my case it took about one minute and was completed.  I later did another [Had need] training session with the [no these] standard text to see what difference it made. (As you can see, it sometimes comes up with strange interpretations of what was said.  I thought I spoke clearly. Pauses tend to cause DNS to insert either a comma or period whether intended by the user or not.)  

The software was now ready for use.  A pop-up window gave me three options, one being to exercise a tutorial, [being to] start dictating, and review new features for experienced users of an older version (if I remember correctly).  I selected the tutorial which took about 28 minutes to complete.  The tutorial presents some general text, a voice presenting a simulated dictation [Dick Tatian] session, and the option to "try it” yourself.  The "try it" sessions allow you to try for yourself whatever you just observed and heard.  I found the "try it" sessions helpful in reinforcing the commands just presented.  When the tutorial was over, the software presented a list of the lessons in the tutorial, allowing the user to return to any individual lesson in practice that lesson.  

At this point, I started dictating this review using Dragon.  Even though there was no lesson explaining how to do this.  I decided to save what I had so far.  The process was very simple.  I merely paused to allow the recognition software to realize I was going to give it a command and stated “File” followed by “save as” and the file name.  

I found in trying to tell the software that I wanted to dictate [activate] the phrase “command file,” it got confused.  When I would say the word “file”, it (DNS) would try to go up to the command line, and activate the drop down menu for the file command,[.]  even [Even] when I was trying to describe it in words.  In this review, I had to resort to manually entering the word “file.”  I've since discovered that if I say the word file quickly with no pause, I can use it (in a text context) and not wind up on the command line.  So this is just a matter of technique and learning to properly use the software.  At my point in understanding the software, I have not been able to adequately install quotes before and after phrases that I thought were quotable (so all quotes you see were manually added after the fact).  So I find that the [it to] keyboard is needed [very needy] at this point in my learning curve to do these little tasks like adding quotes and typing the word file.   

This voice input session ended at 1:12 p.m. July 13.  This exercise was done with no study of the user’s [users] manual, only one pass through the tutorial [provided] (the word “period” was misinterpreted but the following pause produced a period to end the sentence).  The DNS listening software appears to be quite sensitive to pauses and may interpret a pause as a warning that a command will follow.  (When I dictate, I have to stop and think from time to time about what to say next which causes DNS to shift into command mode.  Saying the commands “new line” or “space bar” might have solved this problem.  I suspect with more use, such techniques would become second nature.)}  

(If you are statistically oriented you have probably already counted the number of [] and found there were 17 errors.  There are approximately 1020 words in the text within the curly braces.  That gives an error rate of 1.6 %.  This was by a novice user who had only read the box cover and listened to the tutorial.  Pretty impressive, I thought.)  

There are 66 commands in the index of the User’s Guide that allow the control of the DNS software.  There are 7 commonly used navigation commands some of which are not included in the index.  (This novice user did not study those commands prior to dictating the text above.)  

After reviewing the commonly used voice commands, I tried dictating some quasi legal notes relating to a meeting up in Montana .  Several somewhat legal phrases confused the DNS software.  A great deal of manual editing was necessary to clean up the four page text.  To demonstrate this, a one sentence example was constructed from the phrases that gave DNS the most trouble. This little example was clearly and intentionally loaded against the DNS software to produce a worst case example.  No attempt was made to confuse the software by using strange accents or slurring.  The text was simply read in a normal tone of voice with no pauses.  The example sentence follows:

“The Mountain Harbor Homeowners association wishes to move the original beach easement where the encroachment has occurred to a location where there will be no encroachment.” 

The sentence DNS heard follows:

“[They] Mountain Harbor home[ ]owners Association wishes to move the [old] original beach easement [with] the [coachman] has occurred to a location where [they] [were being on Clinton .]” 

This example shows some of the strange things the DNS hears.  The sample sentence was 25 words long.  DNS misinterpreted 7 out of the 25 or about 28 percent.  DNS has optional special data bases for professions where specialized language is used such as the medical profession.  Perhaps a legal terms data base would have improved the accuracy of the example text.  

Final comments:

• The DNS software is a great improvement over previous voice recognition software I have used in the past.
• Even so, it cannot be trusted to be 100 percent correct.  Some manual cleanup will be necessary and should be expected.
• The installation is straight forward.  Once installed, DNS provides a banner providing pull down menus for user control of the software including turning the microphone on and off.
• When the microphone is turned on, the user interface consists of speaking commands or dictating into a microphone.  Once the commands are mastered, this process should become second nature.
• While it may be my speech patterns that are causing the DNS software to misinterpret what I hear myself say, it is rather disconcerting to have to stop and correct things on the fly.  It rather disrupts the thought processes. 
• The command “scratch that” is supposed to remove the last thing said.  I found that it sometimes behaved too enthusiastically and scratched whole phrases in addition to the last thing said.
• I believe with practice and training, this software will benefit users.  The users will have to learn the commands and how to pace the dictation so the best results can be attained from the DNS software.
• In my case, was it faster for me to deal with correcting DNS then it would have been for me to correct my typing errors?   I don’t know since I never tried typing the whole report separately.  At this stage in my learning curve with this software, it is probably about the same.  I expect with more experience with DNS, DNS will be faster.
• It is kind of amusing to see how a rather smart machine hears your speech and voice patterns. 

Conclusion:  Would I recommend this software to others?   Yes, I would.  The cost is just under $200 for the “Preferred” version that was reviewed.  This provides a unique, powerful, and usable dictation package.  The buyer would have to understand that to get the best use of the DNS software; they will have to learn the commands and the quirks of the software.   That learning commitment should not take more then a few days of use and practice with the software.