Voice file tips? Options

[G_Mark DeVries_*]

Hello everyone,

I know this probably been covered before, but I'm trying to take Dragon NaturallySpeaking more seriously and use it in my work. My current system is a Toshiba M205 tablet PC that has a 1.5 gigahertz Pentium M processor with 1GB of RAM. And thanks to Martin I have a Sennheiser headset with and Andrea USB pod. So hopefully I have an adequate set up. I already started a voice file with the system using Dragon NaturallySpeaking 7.3 medical edition. I choose the large vocabulary and read from the medical reports (harder reading) and then allowed it to scan my documents. Unfortunately, I don't have a lot of documents to scan on this computer. I had begun training some individual words such as new drugs that are out on the market. I try to save my user files at the end of this. I want to try to avoid a corrupt file. Some questions I have are: 1. What size vocabulary would be best? Medium or large? 2. How many passages should I read to train for this file? I have noticed after using the medical reading when I try to train again I don't see another option of a medical reading. Is there another way to access those other medical passages? 3. If I do record notes and save them should I meticulously go through them to take out any errors and save that voice file associated with them and then run the program that allows Dragon to see how you use phrases? I typically have not been saving a voice file that was created with a document. I had planned on after printing out a note to delete the file later. Should I be saving them? How many do I need to have Dragon NaturallySpeaking know me better? 4. I do most my dictation in Microsoft Word 2003 and had adapted some office forms created in Word that uses check boxes, drop-down lists and text fields. These tools have some advantages with a pen because they allow rapid clicking. However, when using voice recognition with the text boxes it is a little harder to select the words individually and Dragon seems a little bit slower. Is this a common problem? I've been playing with settings in Dragon and like the option of double-clicking with the pen and getting the correction box. It seems to be harder when using a text box. 5. How often should back up my voice file? Since I'm just getting started I'm sure I have a lot of personal vocabulary including names of people and drug names I was just curious how often I should do that has a practice once I get it to be working optimally.

Sorry for all the questions, but I do appreciate any opinions or any tips that you might all have. Thanks in advance!

Mark

[Larry Allen]

I'll offer opinions...
1) Since you have few documents to analyze, I suggest the Large vocabulary. When people have LOTS of documents to analyze (as in several months) then I prefer the Medium vocabulary.
2. Don't worry about whether you do training passages using medical readings or Alice in Wonderland. The training analysis analyzes phonemes, not words. If results are good, then I don't suggest any further training... spend the time making corrections, adding phrases, etc. instead.
3. DO save your notes, if for no other reason than for further analysis. Make your needed corrections by keyboard, voice, etc. to make the notes correct for your medical practice.
I've become an advocate of at least trying the Acoustic Optimizer. But that doesn't require using the option to save the audio as other information is saved by default.
4. If using a pen is more productive, use it for those parts of the report. If needed, consider using the "Show Dictation Box" command to bring up a box for dictating longer sections of prose, which is probably where speech will pay off. That sometimes gives better performance than dictating into Word documents.
5. For backup -- for the first month using Dragon I don't worry about backing up the vocabulary files at all. During that time, IMHO it suffices to export your custom words occasionally -- a very tiny file. Why? because during early usage most users change the way they dictate or make other system changes, so restarting with a new user file in case of problems may be just as good as restoring the user file. Besides, DNS automatically makes a backup.
BUT -- you need a backup strategy for your whole system, so during that month that backup strategy should be designed and implemented. You should back up templates, macros, voice files, ... on at least a weekly basis for most medical situations. Sometimes it is better to set it up for daily backup just so it becomes part of a normal routine. Backing up all key files to a CD or USB drive often only takes 10 minutes or so, something easily done during a break, while seeing a patient, or at the end of the day. Do it... I've seen doctors cry or get really cranky when they've lost their data, even if everything is printed and in the charts.

Larry Allen http://www.pcspeak.com

[DAS]

Larry,

Good advice as usual.

As you know, we (Doctors) can get really cranky!

Cheers

DAS

[John Wickett]

Larry,

Of course I would love to disagree with you on a number of points, but there is only one that I think could use more explanation.

I have yet to see any benefit in using the Acoustic Optimizer. I certainly don't see as many problems as in the past, but I just don't feel that it is worth the chance. If someone is going to use it, they should ensure that their voice file is backed up as you described before running the Optimizer.

John

[Larry Allen]

Regarding the Acoustic Optimizer, I've seen it resolve misrecognitions that repeated corrections, making spoken forms, etc. did not resolve.

I fully agree with taking a backup before using the Acoustic Optimizer, as I still don't trust it to leave files in good shape.

Larry Allen
http://www.pcspeak.com

[G_Mark DeVries_*]

Larry, John, and DAS,

Thanks for all your advice. I'll try to implement your suggestions to see if I can get things working optimally.

Thanks guys,

Mark

[Sandi688]

If you are performing additional training texts, but do not run the accoustic optimizer, are there still significant benefits from the additional training?

Thanks,
Sandi

[John Wickett]

Sandi,

>>are there still significant benefits from the additional training?

Not significant, but it helps. However, when doing additional training you are reading text that is likely quite different than your normal speech patterns and verbiage. I personally feel that properly correcting your own dictation yields better accuracy in the end result.

John