KnowBrainer Speech Recognition
Decrease font size
Increase font size
Topic Title: Vocabulary
Topic Summary: How the vocabularies work, active, inactive/backup and custom
Created On: 02/22/2012 05:30 PM
Status: Post and Reply
Linear : Threading : Single : Branch
 Vocabulary   - DessaVan - 02/22/2012 05:30 PM  
 Vocabulary   - R. Wilke - 02/22/2012 05:52 PM  
 Vocabulary   - DessaVan - 03/11/2012 06:15 PM  
 Vocabulary   - R. Wilke - 03/11/2012 06:52 PM  
 Vocabulary   - DessaVan - 03/11/2012 07:16 PM  
Keyword
 02/22/2012 05:30 PM
User is offline View Users Profile Print this message


DessaVan
Power Member

Posts: 48
Joined: 05/11/2011

Dear Dragon Braniacs,

I teach Dragon to court reporters and often find myself in a discussion about whether or not to do corrections, how to do corrections and how the vocabularies work.

I do corrections the way Chuck recommended; I correct utterances rather than highlighting and correcting single words.  

Court reporters have been using Dragon (and ViaVoice) for over ten years now and there are many ideas about how the vocabulary works.

Over the years I have heard different ideas about the vocabularies.  There are thoughts that when you add a word/phrase to the custom vocabulary, you must go into the active vocabulary and delete the equal number of words.  If you don't delete any words, Dragon will do it and may delete a word you will need such as "the" or words that we need for everyday speech.  There is a "fear" that adding phrases to the custom vocabulary will make active words somehow unavailable, so there is a reluctance to build up the custom vocabulary. (I don't believe this to be the case.)

I've had court reporters tell me they spend hours deleting words from the active vocabulary to improve their recognition.   

I think the "active" vocabulary contains the words the user uses most often.  Words we don't use move to the inactive vocabulary until and if we need them.  I want the people I train to have success with Dragon so I'm here or the other forum trying to collect the most accurate information.

My understanding is that while a word may roll to the inactive vocabulary, it is still available if we need it. 

Will analyzing documents bring words up from the inactive dictionary?

Will Dragon pull a word from the inactive vocabulary if we just say it? 

I see many words in my vocabulary with green up arrows.  I know I have never spoken any of those words, I didn't add them, yet they are there in my active vocabulary.  Is it because Dragon considered that word for a choice? 

Does adding words to a custom vocabulary remove words from the active vocabulary?

I thought that I read somewhere that Dragon had the capacity for something like 150,000 custom words/phrases without any impact on the active vocabulary.

I have always taught that Dragon can't give you a word it doesn't have...Lunis?  Dessa?  I add lots of word combinations to my custom vocabulary so they will translate the way I want them to and I don't have to correct them all the time. 

Waffle House is an example.  I get waffle House when I dictate with a new user but when I add Waffle House to my custom vocabulary, it comes up perfectly every time.  

One more question, if you know, how do the Dragon vocabularies work compared to ViaVoice vocabularies? 

I so appreciate the time you all put into answering long questions like mine.

Thanks, guys.

Dessa 

 02/22/2012 05:52 PM
User is offline View Users Profile Print this message

Author Icon
R. Wilke
Top-Tier Member

Posts: 4391
Joined: 03/04/2007

Quote:
Over the years I have heard different ideas about the vocabularies. There are thoughts that when you add a word/phrase to the custom vocabulary, you must go into the active vocabulary and delete the equal number of words. If you don't delete any words, Dragon will do it and may delete a word you will need such as "the" or words that we need for everyday speech. There is a "fear" that adding phrases to the custom vocabulary will make active words somehow unavailable, so there is a reluctance to build up the custom vocabulary. (I don't believe this to be the case.)

There is no reason for such fear. Starting with version 10, you can add about 150,000 custom words (about 50,000 in version 9) before you hit the ceiling.

Quote:
I've had court reporters tell me they spend hours deleting words from the active vocabulary to improve their recognition.

Pretty much useless.

Quote:
I think the "active" vocabulary contains the words the user uses most often. Words we don't use move to the inactive vocabulary until and if we need them. I want the people I train to have success with Dragon so I'm here or the other forum trying to collect the most accurate information.

Not really close to it. The "active" vocabulary comes with about 152,000 words from the start, when using the general large base topic. It doesn't matter whether or not you use any of them for them to stick around, unless you actively delete them or hit the ceiling per the above (highly unlikely considering the sheer volume available).

Quote:
My understanding is that while a word may roll to the inactive vocabulary, it is still available if we need it.

It won't "roll" there by itself, but if you delete it it will still be available - that is, you cannot change the backup dictionary (which you are referring to as "inactive", it is hard coded altogether.

Quote:
Will analyzing documents bring words up from the inactive dictionary?

Yes.

Quote:
Will Dragon pull a word from the inactive vocabulary if we just say it?

No, because they are not active. A word will be retrieved from the backup dictionary under any of three circumstances, (a) correction, (b) spelling, and (c) analyzing documents for new words.

Quote:
I see many words in my vocabulary with green up arrows. I know I have never spoken any of those words, I didn't add them, yet they are there in my active vocabulary. Is it because Dragon considered that word for a choice?

No, not because of that, but maybe you did one of the above. Also, if you delete a word from the (initially) active vocabulary, and then add it back any time later, it will carry a green arrow from then on.

Quote:
Does adding words to a custom vocabulary remove words from the active vocabulary?

Only if and when you hit the ceiling.

Quote:
I have always taught that Dragon can't give you a word it doesn't have...Lunis?

That's right, Dragon doesn't understand Lunis natively.

Rüdiger

 



-------------------------

Well, it's past the point where we can make any changes in the code, but we can still make changes to the Easter Egg!

 03/11/2012 06:15 PM
User is offline View Users Profile Print this message


DessaVan
Power Member

Posts: 48
Joined: 05/11/2011

Thank you.

I have looked at the Nuance Web site as well as Google to find concrete info about Dragon 11's vocabulary size, active and custom.

I can't find it anywhere.  I need to cite a reliable source.  You'd think it would be posted somewhere and I just can't find it.

Lunis mentioned a vocab size of "approximately 170,000" while you mention about 152,000 active and about 150,000 custom.

While your information is valuable, there is a discrepancy between your numbers and Lunis'.

What do you base your info on?

Is there a place to find these numbers?

Chuck?

Thanks again for your time.

I learn so much going through these forums.

Dessa

 03/11/2012 06:52 PM
User is offline View Users Profile Print this message

Author Icon
R. Wilke
Top-Tier Member

Posts: 4391
Joined: 03/04/2007

Quote:
While your information is valuable, there is a discrepancy between your numbers and Lunis'. What do you base your info on? Is there a place to find these numbers? Chuck?

Now, that's interesting. I base my info on knowledge retrieved from using the SDK, as opposed to the other sources you make reference to. But you don't have to take my word for it, at least in parts. Open the vocabulary, and make sure you select the entire list, then click the "Train" button. A message will display telling you that the amount of words you have selected for training is not legal, but also inform you how many words you have actually selected.

As far as the room in the vocabulary reserved for adding words, that's a property you can retrieve from the SDK, if you know how to. Like I said, since version 10 the number has been around 150,000.

Rüdiger

 



-------------------------

Well, it's past the point where we can make any changes in the code, but we can still make changes to the Easter Egg!

 03/11/2012 07:16 PM
User is offline View Users Profile Print this message


DessaVan
Power Member

Posts: 48
Joined: 05/11/2011

I did the train words trick.

150980 words in the active vocabulary, thank you very much.

 

 

KnowBrainer Speech Recognition » NaturallySpeaking Speech Recognition » Vocabulary

Statistics
27371 users are registered to the KnowBrainer Speech Recognition forum.
There are currently 5 users logged in.
The most users ever online was 2028 on 04/05/2013 at 07:36 PM.
There are currently 146 guests browsing this forum, which makes a total of 151 users using this forum.

FuseTalk Standard Edition v4.0 - © 1999-2013 FuseTalk™ Inc. All rights reserved.