Windows 7 for hands free computing?
Printed From: MSSpeech-Forum
Category: Windows™ Speech Recognition Forums
Forum Name: Accessibility for the Disabled
Forum Description: Using WSR to Interface with the computer by voice
URL: https://www.msspeech-forum.com/forum_posts.asp?TID=76
Printed Date: 24/Dec/2024 at 11:46pm Software Version: Web Wiz Forums 12.02 - http://www.webwizforums.com
Topic: Windows 7 for hands free computing?
Posted By: jnuttallphd
Subject: Windows 7 for hands free computing?
Date Posted: 21/Apr/2009 at 8:53am
I am currently a Dragon NaturallySpeaking user. I have ALS and am lossing some control over my hands. I will need to be able to control the computer hands free. I am thinking of getting a new computer when Windows 7 becomes available. I'm wondering with Windows 7 and the WSRToolKit will I be able to do hands free computing. I am looking forward to WSR in Win 7. So anyone with any experience with WSR in Vista or Win 7 your comments on hands free will be greatly appreciated.
Jim
|
Replies:
Posted By: mmarkoe
Date Posted: 21/Apr/2009 at 11:01am
Jim,
See http://www.msspeech-forum.com/forum_posts.asp?TID=59&title=whats-new-in-wsr-for-windows-7 - What's New in WSR for Windows 7 .
The reports we are getting from beta testers is WSR and the WSRToolkit run just fine in Windows 7. Although we have not tested it in Windows 7, in previous versions, you could have both Dragon Naturally Speaking and WSR open at the same time. Just make sure the microphone is only actively on in one of them at a time.
We feel that command and control/accessibility is greater in WSR. Have you seen or tried the "Show Numbers command?" It is very useful for navigation.
There's a vast library of macros. If you're skilled at creating macros or would like to learn how, consider Brad Trott's PDF book, " http://www.mymsspeech.com/microphones/prod_details.asp?prodID=231 - WSRMacros: The User's Guide ." The WSRToolkit at $15.99 and the Macro Guide at $9.99 may be the best value in computer software you will ever purchase.
Marty
------------- Marty Markoe, MVP
Microsoft Valued Partner
See us at: http://www.mymsspeech.com
|
Posted By: jnuttallphd
Date Posted: 21/Apr/2009 at 11:56am
Hello Marty. Thanks so much for the great response. Currently I use Windows XP. So I have not had a chance to try WSR. In the demonstrations I have seen from video clips on the Internet WSR looks quite good. I like the fact that it is integrated into the operating system. I think this integration has a advantage for command and control, which is what I am looking for. I do like to write my own macros. I have Dragon NaturallySpeaking 10 professional. This program was purchased through my work. Since I am retiring I will not be able to upgrade my DNS program in the future. So I'm looking for a good alternative. I have read the manual for the WSRTools. This looks like a great way to be able to create your macros. It looks very similar to what I've been doing with DNS. So again thanks for the help. I'm really pleased at the WSRTools works in Windows 7. Any other insights you can provide will be greatly appreciated. Jim
|
Posted By: mmarkoe_admin
Date Posted: 21/Apr/2009 at 1:42pm
Jim,
The WSRToolkit is free to people with disabilities. You qualify. Please send an E-mail requesting it when you get Windows 7 up and running.
One thing I've noticed with WSR is it is much less forgiving of the slurred speech. Just try to enunciate every word as clearly as possible and you will be up and running very quickly.
Marty
|
Posted By: jnuttallphd
Date Posted: 21/Apr/2009 at 2:04pm
Posted By: Jerry
Date Posted: 21/Apr/2009 at 9:29pm
Hello Jim. Welcome to the forum.
Many of us are deciding to wait for Windows 7 before we purchase a new computer. It is a good idea as the release of Windows 7 especially for new computers is not too far away. In the design of your computer there are a number of options that have really never been available before. The most significant one is the choice between a 32-bit operating system and a 64-bit operating system. In my experience, the amount of available RAM directly impacts on voice dictation performance. A 32-bit operating system whether it be Windows 7 or Vista allows a maximum of 4 gigs of RAM. A 64-bit operating system, again, Windows 7 or Vista allows a huge amount of memory to be available. When I upgraded my 32 bit Vista ultimate system from 2 gigs to 4 gigs of memory I found a huge difference particularly in speed and recognition quality. I have found this to be true for both DNS 10 and WSR using Vista. Until very recently, DNS 10 was not available for a 64-bit operating system. Today it is and is a free upgrade to DNS 10.1, that is if you are using DNS 10. You have indicated that you are so you will be free to use either operating system utilizing either 32 bits or 64-bits.
Like Marty has discussed before on this forum, one of the key ingredients to improving voice dictation accuracy is the quality of the microphone. I have also experimented and experienced significant improvement when using a USB microphone vs. a microphone that plugs into a soundcard. I have a very good soundcard, Creative Labs Audigy Platinum, and the DSN 10 sound checks average between 19 and 21. Using a USB plug I get between 23 and 26 regularly on the same microphone. The accuracy level reflects this difference. Another consideration will be whether to use a wired microphone or a wireless microphone. I am a power wheelchair user and really enjoy the freedom that the wireless microphone gives me. I use the Samson air 77 and find it to be quite outstanding. As I wear the headset for about 10 hours a day, I had to fashion some cushioning to help prevent soreness around the ears but it works so well that it is worth it. Marty has some really good information on it on his website.
Deciding upon which voice dictation system to use becomes a personal choice and you might alternate depending on the task you're trying to accomplish. Personally, I use both and find that for dictation quality I get better results with DNS 10. However, for control of the Windows environment, WSR tends to be a little easier. I look forward to hearing about your experiences.
------------- Take care, Jerry
|
Posted By: cmwilkerson
Date Posted: 05/May/2009 at 10:28am
I've noticed that in Win 7, with WSR, (open but off) and I open a pdf file, WSR offers to scan it for me. Presumably the scan is for word recognition. I wish I could do this with my Word documents and rtf files.
Anybody have any experience with this? I have converted (cutepdf) some Word docs to pdf for scanning, but I would rather not have to do this.
------------- Dr. Chris Wilkerson
Carson Doctors Group
MVP TabletPC
www.MedicalTabletPC.com
www.Digital-Doc.com
|
Posted By: mmarkoe
Date Posted: 05/May/2009 at 10:55am
cmwilkerson wrote:
I've noticed that in Win 7, with WSR, (open but off) and I open a pdf file, WSR offers to scan it for me. Presumably the scan is for word recognition. I wish I could do this with my Word documents and rtf files. |
The http://www.mymsspeech.com/microphones/prod_details.asp?prodID=228 - WSRToolkit (see image of Interface below my name) has this feature and much more.
Add From File allows you to improve the accuracy of your "Language Model". It's not the same as adding words to your personal dictionary but works by taking the text that you pass it, parses it into words, and then records the relative frequencies of occurrence of each word with respect to the words around it. If you use documents that represent your style of speaking, this feature will improve dictation accuracy.
Train From Text purpose is is to improve the acoustic model of your user profile. The acoustic model is your specific way you enunciate words. We recommend you start with the default text, "The Rainbow Passage." Then read a few of your typical documents.
By running the two utilities above you are likely to notice an immediate improvement in accuracy.
Marty
------------- Marty Markoe, MVP
Microsoft Valued Partner
See us at: http://www.mymsspeech.com
|
Posted By: cmwilkerson
Date Posted: 05/May/2009 at 11:03am
Thanks Marty,
I will run the toolkit as well. My biggest fear is bloating my profile to the point of recognition accuracy declining. Any advice on this? Is there a sweet spot?
------------- Dr. Chris Wilkerson
Carson Doctors Group
MVP TabletPC
www.MedicalTabletPC.com
www.Digital-Doc.com
|
Posted By: mmarkoe
Date Posted: 05/May/2009 at 12:20pm
cmwilkerson wrote:
I will run the toolkit as well. My biggest fear is bloating my profile to the point of recognition accuracy declining. Any advice on this? Is there a sweet spot? |
WSR does not seem to be a resource hog as compared to Dragon Naturally Speaking. We have a notebook computer and a desktop at the office that have user profiles that are 2 1/2 old and I cannot discern any performance slow down with these machines as compared to any machines with newer profiles. These mature user profiles have many words, phrases, and macros added as well as having run documents through the add from file and train from text features.
Marty
------------- Marty Markoe, MVP
Microsoft Valued Partner
See us at: http://www.mymsspeech.com
|
Posted By: cmwilkerson
Date Posted: 05/May/2009 at 12:44pm
mmarkoe wrote:
cmwilkerson wrote:
I will run the toolkit as well. My biggest fear is bloating my profile to the point of recognition accuracy declining. Any advice on this? Is there a sweet spot? |
WSR does not seem to be a resource hog as compared to Dragon Naturally Speaking. We have a notebook computer and a desktop at the office that have user profiles that are 2 1/2 old and I cannot discern any performance slow down with these machines as compared to any machines with newer profiles. These mature user profiles have many words, phrases, and macros added as well as having run documents through the add from file and train from text features.
Marty |
Marty,
I think you misunderstood my question. My fault. I am not referncing, "Resources," in the sense of CPU cylces and or memory. I am referring to a scenario of say having a Lexion of 100,0000 words vs 10,000 words and the affect on recognition.
------------- Dr. Chris Wilkerson
Carson Doctors Group
MVP TabletPC
www.MedicalTabletPC.com
www.Digital-Doc.com
|
Posted By: mmarkoe
Date Posted: 05/May/2009 at 5:37pm
[/QUOTE=cmwilkerson]I think you misunderstood my question. My fault. I am not referncing, "Resources," in the sense of CPU cylces and or memory. I am referring to a scenario of say having a Lexion of 100,0000 words vs 10,000 words and the affect on recognition. [/QUOTE]
In Dragon there's a lexicon of approximately 160,000 words from the English language. However, there is a limit to what is in the actual active vocabulary. I believe for Dragon it is something like 10,000 words. Therefore, almost all the words you think you were adding are actually already in the lexicon. When you correct a word or add a new word not in the lexicon, you're actually just moving it into the active 10,000 word lexicon and something else that has not been used is moved out of the active vocabulary.
I would guess that WSR Works the same way. If the active vocabulary were so large, it would really slow things down completely. I think when Dragon starts to choke it is because of many incorrectly corrected words and nonsense words that get into the system.
Marty
------------- Marty Markoe, MVP
Microsoft Valued Partner
See us at: http://www.mymsspeech.com
|
Posted By: cmwilkerson
Date Posted: 09/May/2009 at 12:06pm
Marty,
Thanks for the reply.
I'm using WSR on Windows 7 with a Lenovo Tabletpc X200 with internal array mic and 4 gigs of RAM. It works great but needs a medical lexicon. For example, the word, "radiculitis," is seen as "ridiculous." Using the Tablet Dictionary's import tool, I imported 1000 medical words.
After the import, radiculitis is recognized, as well every medical word tested that resides in my import!!!
At this point, for my needs, there is no need for me to reinstall DNS.
------------- Dr. Chris Wilkerson
Carson Doctors Group
MVP TabletPC
www.MedicalTabletPC.com
www.Digital-Doc.com
|
Posted By: mmarkoe
Date Posted: 09/May/2009 at 5:37pm
cmwilkerson wrote:
After the import, radiculitis is recognized, as well every medical word tested that resides in my import!!! At this point, for my needs, there is no need for me to reinstall DNS. |
That's a great endorsement for the WSRToolkit. Do you mind if I mention this thread and your report in other forums?
Thanks, Marty
------------- Marty Markoe, MVP
Microsoft Valued Partner
See us at: http://www.mymsspeech.com
|
Posted By: cmwilkerson
Date Posted: 09/May/2009 at 6:54pm
mmarkoe wrote:
cmwilkerson wrote:
After the import, radiculitis is recognized, as well every medical word tested that resides in my import!!! At this point, for my needs, there is no need for me to reinstall DNS. |
That's a great endorsement for the WSRToolkit. Do you mind if I mention this thread and your report in other forums?
Thanks, Marty |
Hi Marty,
I was using the Dictionary Tool by Microsoft, a free powertoy to import a complete list of words. The WSR Toolkit, to my knowledge only allows me to add individual words. I do use the WSR Toolkit to Add from File which then scans various charts. That way, I think I get both the Lexicon and the language model.
------------- Dr. Chris Wilkerson
Carson Doctors Group
MVP TabletPC
www.MedicalTabletPC.com
www.Digital-Doc.com
|
Posted By: mmarkoe
Date Posted: 10/May/2009 at 12:01am
cmwilkerson wrote:
I was using the Dictionary Tool by Microsoft, a free powertoy to import a complete list of words. The WSR Toolkit, to my knowledge only allows me to add individual words. I do use the WSR Toolkit to Add from File which then scans various charts. That way, I think I get both the Lexicon and the language model. |
You should take a look at all the features in the WSRToolkit. What you are calling the Dictionary Tool is our Add From File mentioned in our response to you Posted: 05/May/2009 at 10:55am. I also mentioned Train From Text. These two work nicely together one improves Language Model and the other the Acoustic Model. You are unlikely adding words as the words are already there. By running Add From File you are bringing the words to the front of the Language Model by increasing their frequency of usage. See the features of the WSRToolkit:
Add From File allows you to improve the accuracy of your "Language Model". It's not the same as adding words to your personal dictionary but works by taking the text that you pass it, parses it into words, and then records the relative frequencies of occurrence of each word with respect to the words around it. If you use documents that represent your style of speaking, this feature will improve dictation accuracy.
Train From Text purpose is is to improve the acoustic model of your user profile. The acoustic model is your specific way you enunciate words. We recommend you start with the default text, "The Rainbow Passage." Then read a few of your typical documents.
Marty
------------- Marty Markoe, MVP
Microsoft Valued Partner
See us at: http://www.mymsspeech.com
|
Posted By: cmwilkerson
Date Posted: 10/May/2009 at 12:11am
>>What you are calling the Dictionary Tool is our Add From File mentioned in our response to you<<
No, it is not. I am referring to a powertoy from Microsoft that is called, "dicitonarytool.exe that allows me to import an entire collection of medical words that empirically improve my recognition result.
------------- Dr. Chris Wilkerson
Carson Doctors Group
MVP TabletPC
www.MedicalTabletPC.com
www.Digital-Doc.com
|
Posted By: mmarkoe
Date Posted: 10/May/2009 at 9:02am
OK, now I see you are referring to the tool downloadable for Tablet PCs at:
http://www.microsoft.com/windowsxp/downloads/powertoys/tabletpc.mspx - http://www.microsoft.com/windowsxp/downloads/powertoys/tabletpc.mspx
I did not realize this works for WSR as it was designed for XP Tablet.
What you are doing is adding words but not the acoustic information. The acoustic information is added when those same words would be trained using the Train From Text feature of the WSRToolkit. You said you were using the Add From File feature and that does the same thing as the dictionary tool. For medical words, the acoustic trainining may not be as critical since the words used are so unique in their pronounciations.
I believe the dictionary tool was designed for handwriting recognition on Table PCs. As the handwriting and SR software use the same dictionary this would work well for medical vocabulary.
Marty
------------- Marty Markoe, MVP
Microsoft Valued Partner
See us at: http://www.mymsspeech.com
|
Posted By: cmwilkerson
Date Posted: 10/May/2009 at 10:18am
>>As the handwriting and SR software use the same dictionary this would work well for medical vocabulary.<<
Exactly why I used Microsoft powertoy dictionayrtool.exe. I have some more specific instructions, with respect to the tool, found here: http://www.medicaltabletpc.com/index.php?option=com_content&task=view&id=660&Itemid=1 - http://www.medicaltabletpc.com/index.php?option=com_content&task=view&id=660&Itemid=1
------------- Dr. Chris Wilkerson
Carson Doctors Group
MVP TabletPC
www.MedicalTabletPC.com
www.Digital-Doc.com
|
Posted By: mmarkoe
Date Posted: 10/May/2009 at 10:27am
[QUOTE=cmwilkerson]Exactly why I used Microsoft powertoy dictionayrtool.exe. http://www.medicaltabletpc.com/index.php?option=com_content&task=view&id=660&Itemid=1%5b/QUOTE - [/QUOTE ]
You are obviously much more highly technically proficient than most physicians. We posted on the http://www.emrupdate.com/forums/p/19481/97356.aspx#97356 - EMRUpdate forum how using the WSRTooklit does the same thing you do but has the tools readily available. Also, I believe you increase accuracy by using both Add From File and Train From Text.
Marty
------------- Marty Markoe, MVP
Microsoft Valued Partner
See us at: http://www.mymsspeech.com
|
|