Apologies if this topic appears twice, IE crashed on me the first tiem I submitted it

slayer,

I didn't see your post appear anywhere else. What is your question regarding dra files?

Chuck Runquist
Technical Project Manager
VoiceTeach LLC
Home of VoicePower®: We don't make Dragon NaturallySpeaking, We make it better!

"We are all faced with a series of great opportunities brilliantly disguised as insoluble problems." - John W. Gardner

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

Not having a good day.

I wanted to ask what the dra files in the drafiles folder of the voice_container folder were and how they get created. I ask because a number of my colleagues have Dragon profiles that have grown quite large ( > 1GB - non-roaming by the way) in a relatively short space of time. The main contributer seems to be the DRA files. The names of these files imply to me that they are being created automatically rather than by the user. Is there a tick box I have missed that will stop them being created or a limit I can put on the folder size? Is there a better way to get rid of them other than deleting them iin windows explorer from both current and backup or will this has nasty ramifications? I'm not a Dragon user myself.

Having wanted to get rid of them it may seem strange but can I also ask is there any way to combine dra files ie append one onto another?

Thanks

slayer,

Let me answer your last question first. No, there is no way to append one onto another. Attempting to do so would be a disaster. There are other ways of handling dra files in the drafiles folder.

First, dra files are Dragon's or prior to reformat "Dragon Recorded Document (*.dra). These files are created automatically as part of the SilentAdapt feature. Every time you dictate anything, and upon completion of that particular dictation, Dragon records such in the drafiles folder. Yes, they can get very large because these files contain both the audio and the text combined for each document dictated. The reason the files can be so large is because every minute of dictation stores approximately 1.5 MB of audio data.

Second, these files are useful because they are used by the Acoustic and Language Model Optimizer (Accuracy Tuning) to adapt to your user profile. When the Acoustic and Language Model Optimizer acts on these files, it adapts the audio portion similar to the adaptation process that occurs when you run a General training. It adapts the text portion similar to running the "Learn from specific documents" in the Accuracy Center. This can be very useful in terms of improving the overall accuracy of DNS. When you run the Acoustic and Language Model Optimizer, DNS uses these files to adapt to your user profile, after which it consolidates and deletes them.

If you don't run the Acoustic and Language Model Optimizer, or if you choose not to, these dra files will continue to accumulate, and thus significantly increase the size of your user profile. If you run the Acoustic and Language Model Optimizer every few days, this will consolidate and reduce the overall size of your user profile. Some users think it not useful to do so. Some users, including and especially myself, find that it tends to improve overall accuracy. The choice is up to the user. It does no harm to run the Acoustic and Language Model Optimizer, whether or not you get improved accuracy and performance by doing so, but it also does no harm not to, except as noted with regard to increasing the user profile size by virtue of the accumulation of these files.

Third, since you have the Professional version, you can see what these files contain by doing the following:

1.  Open DragonPad.

2.  Select Open from the File menu or click on the folder icon in the toolbar.

3.  Browse or go to the drafiles folder. Open the drop-down list at the bottom right [i.e., the default is Rich Text Format (*.rtf)], and select "Dragon Recorded Document (*.dra). You will see the dra files listed.

4.  Select one and open it. In all likelihood you will recognize what you dictated from the text contained therein.

5.  Select some text and say "play that back". You will hear the audio behind the text.

Note that this is only available to Professional version users, so those who have Premium and below can take advantage of this feature.

Also, this is sometimes very useful if you lose a document. If Dragon has recorded such in the dra file in this folder, you can recover it. However, it's not easy to find what you're looking for because Dragon doesn't store these in any kind of order that makes it easy to search. Regardless, it's a very useful feature and it has saved my butt on any number of occasions.

On the other hand, if you do not intend to use the Acoustic and Language Model Optimizer and you do not want to store these files in the drafiles folder, you can turn this feature off very simply by going to the Dragon Options dialog | Data tab and adding a checkmark to "Conserve disk space required by user profile (for portability)". This will effectively eliminate the creation and storing of dra files in this folder. After you do that, you can go to the drafiles folder and delete all the content (dra files). You should also go to the voice_container folder and delete the "drafiles.ini" file. This will prevent any untoward errors from occurring by virtue of simply deleting the dra files in the drafiles folder.

It away, your choice. But if you intend to keep these files, then be sure that you run the Acoustic and Language Model Optimizer regularly so that these files don't accumulate making your user profile exceptionally large.

Chuck Runquist
Technical Project Manager
VoiceTeach LLC
Home of VoicePower®: We don't make Dragon NaturallySpeaking, We make it better!

Education is when you read the fine print. Experience is what you get if you don't. - Pete Seeger

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

Thanks very much for such a swift, clear and concise answer. I'll pass your suggestions on.

Chuck,

it's just the opposite. You need to set the checkmark in order not to save the DRA files, therefore conserve (save) disk space.

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!

Rüdiger,

Thank you for catching that. Sometimes my dyslexia kicks in, particularly if I put my hat on backwards.

You are absolutely correct and I screwed up.

Chuck Runquist
Technical Project Manager
VoiceTeach LLC
Home of VoicePower®: We don't make Dragon NaturallySpeaking, We make it better!

"The best of us must sometimes eat our words." - J. K. Rowling

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

I need to ask your advice again about these dra files. We are now trying to use roaming profiles. We have ticked the box on the Tools/Options misclleaneous tab which says 'When the User Profiles Closes' -  Automatically save the profile changes. We hope that means that whenever we close the profile, the master roaming profile is updated with changes. Just getting to the DRA files question. We have found that newly created dra files in the local cahed copy of the roaming profile are not being copied to the master roaming profile on our server. We have tried two different profiles on two different pcs with the same result. I don't understand why not. I can see the timestamp change on the master copy on the server so I know the profile has been updated yet no dra files. Why? Thanks iin advance.

Slayer,

 

Absolutely correct. I suggest you tick the "vocdelta" box too (I don't have the proper description handy in English right now). The vocdelta.dat is the file that actually transfers the changes in a locally cached user profile to the master server profile. Ticking the box should speed up the synchronisation process. 

Anyway, when I set up a roaming user installation, I usually tick boxes No. 3-7 and limit archive size on the network to 1 GB.

Don't worry about it. The locally cached dra files will be transferred to the server folder in due time when the local archive has reached a certain size. You don't need them anyway except for acoustic optimisation, the relative merits of which have been discussed frequently on this forum. 

Stephan

 

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

www.egs-vertrieb.de - Speech Recognition Blog - Forum: www.immer-eine-Nuance-besser.de

Thanks for that information. I forsee a problem then in that our users want to keep their dra files but as there may be quite a few different people using the pcs, the aim was to delete the local cached copy of their roaming profile when finished for the day for housekeeping purposes in order to avoid filling up the local hard drive. Our locally cached dra files have never made it up to master copies on the server by the way.

 Can I also ask about the non-roaming to roaming conversion. We are currently using the Advanced/Save to roaming feature in Manage profiles which is very time consuming as we have quite a few users, most of which have quite a few dra files in their profile. However, the  help file I am reading says you can just copy and paste user folders from the non-roaming profiles in C:\Documents and Settings\All Users\Application Data\Nuance\NaturallySpeaking11\Users and use these as the master roaming copy. Any we have converted by the Save to Roaming  method contain current and tb folders plus the file audioin.dat, so is that all I need copy from the non-roaming version which to be honest would be a much quicker way to convert everyone  in one go and then we wouldn't have to rely on the users themselves to do the converting? Obviously we don't need the backup folders and I surmise that the chkpt folder contians 'checkpoints' which can be used to retrieve a profile if both the current and backup profiles are corrupt.  I wondered though what the dataarchive folder is and why it isn't needed in a non-roaming profile. Any I looked in seem to be empty anyway.

 Lastly, I'm not very sure about the 'archive' at all and what it is, in general terms.

slayer,

Whatever you do don't play around with copying only portions of the user profile. You will definitely screw up your roaming users. Either copy the complete user profile folder or use the Advanced | Save to roaming option. Yes, it's slow, but it won't screw up your user profile(s). By design, Dragon provides these kinds of features specifically to avoid messing up your user profile(s)

Also, the chkept folder does not do what you seem to think it does. That is, it won't retrieve a good user profile if both the primary and backup users are corrupted. That's not its purpose. Regardless, don't play around with that folder.

I would suggest that you go back through the Dragon Help and thoroughly review the sections on setting up and using Roaming user profiles. I think you will find a good deal of your questions answered there, particularly with regard to how Dragon handles dra files Re: Roaming users.

Many of the folders like the dataarchive folder and the tb folder are temporary. That is, once they have served their purpose, the folder is emptied (blank). Again, leave these folders alone because Dragon takes care of them for you anyway.

As regards the backup folders, only the administrator on the network can back up the Master Roaming user profiles. However, if you want to eliminate backup folders in the non-roaming user profiles, just set that option in the DNS Options dialog | Backup user profiles every [ ] saves, just set that entry to blank (no entry).

Again, if you carefully review all the information related Roaming users in the Dragon Help, I think you will find virtually all of your answers there. It's pretty comprehensive, if a bit confusing in some places.

Chuck Runquist
Technical Project Manager
VoiceTeach LLC
Home of VoicePower®: We don't make Dragon NaturallySpeaking, We make it better!

"Life's Rule #1: Once you pull the pin, Mr. Grenade is no longer your friend."  (Variant of Murphy's Law  -  Edward A. Murphy, Jr)

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

Slayer,

The dra files in the local archive are good for absolutely nothing. Anyone who wants to keep them - well, serves them right. (Do not confuse them with the recordings asssociated with dictated documents, which also have the format .dra - but these are not stored in the user profile.)

If and when the files are transferred to the server for the purpose of acoustic optimisation, the mighty Dragon alone knows. Acoustic optimisation is something most user can do very well without, and may even harm accuracy. Again, anyone   who wants to keep them - well, serves them right. Feel free to use your housekeeping utilities to get rid of them regularly.

Do not delete the locally cached copies of the user files! These are the files and folders that Dragon actually works with. If you delete them overnight, users will have to download their full profile from the server in the morning. Not a good idea.

Absolutely. You don't need to copy the dra files in the voice_container folder, either.

The only small caveat is the roaming.ver file that contains some encrypted information about the Roaming user. I have never come across any issues when this file isn't there, but I can't rule it out completely. BTW I usually set up Dragon in a way that all users are roaming users by default. Saves the conversion.

The chkpt folder testifies to Nuance's intention of creating said checkpoints. They haven't managed that yet, though. Look out for that feature in V. 12 or later.

Nobody knows what the dataarchive folder is for.

 

Again, it's where files are stored for acoustic optimisation, should anybody want it. I don't. Empty it at regular intervals.

Stephan

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

www.egs-vertrieb.de - Speech Recognition Blog - Forum: www.immer-eine-Nuance-besser.de

 Stephan, I'm ashamed to say I don't know where the archive is to be found.

Stephan,

I have only one comment. If you believe what you say, then I have two bridges in New York City that I would love to sell you. Unfortunately, your statements simply are not correct.

There are times when you run the Acoustic and Language Model Optimizer where you may not notice any improvement in overall accuracy. However, every time I run the Acoustic and Language Model Optimizer I see improvements in some contexts. Once you get to 99.9% accuracy overall and 100% accuracy much of the time, that means that you are getting one misrecognition for every 100 words dictated. That's as good as it gets. Most of the misrecognitions that I see are related to words that have similar pronunciations, such as two, too, and to or and, end etc. Even so, running the Acoustic and Language Model Optimizer often improves the accuracy in contexts that use these words.

Regardless of how you view the functionality of the Acoustic and Language Model Optimizer, passing that on to end-users as an "absolute" simply isn't providing any useful information. Let users decide for themselves whether or not dra files and/or running the Acoustic and Language Model Optimizer are beneficial. Retaining dra files or running the Acoustic and Language Model Optimizer doesn't do any damage to your user profiles or make them any less accurate.

I totally disagree with your assessment in this case.

Chuck Runquist
Technical Project Manager
VoiceTeach LLC
Home of VoicePower®: We don't make Dragon NaturallySpeaking, We make it better!

You can't convince a believer of anything; for their belief is not based on evidence, it's based on a deep seated need to believe. -  Dr. Carl Sagan

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

Chuck,

I heard that some Americans have bought London Bridge because they thought it was Tower Bridge and were quite disappointed when they didn't see the Gothic ornaments :-)

Of course, you're absolutely right to say that the Acoustic Optimizer is a matter for dispute, and that it may do good to some. I apologise if I have been too harsh in my judgement.

What I have found in my personal practice, which mainly involves supporting Dragon NaturallySpeaking resellers, is that some customers insist on running the Acoustic Optimizer every so often and causing all sorts of trouble on the network. I cannot judge whether this is due to a faulty setup, sloppy dictation habits or whatever, but it is one very good reason for me not to recommend the Acoustic Optimizer.

On a more theoretical level, I still cannot see why the Acoustic Optimizer should do much good if Dragon keeps updating and adapting the profile during the process of dictation anyway. But this has been discussed many times without any conclusive evidence or answer, including this thread, and including a discussion of whether running the Acoustic Optimizer decreases or increases the size of the user profile.

Shall we just agree to disagree,

Best wishes, Stephan

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

www.egs-vertrieb.de - Speech Recognition Blog - Forum: www.immer-eine-Nuance-besser.de

Stephan,

I've never found that there are any limitations on poor logic based on country, culture, group or related factors. Consider one of the famous quotes by Albert Einstein.
"The world we have created is a product of our thinking; it cannot be changed without changing our thinking."

My point is only this; one of the first rules of technical support is that you don't recommend that users take any particular approach, especially if it's based on your own personal experiences or opinions. That is simply closing the door on options. When you tell a customer not to do something and they consider you to be an expert, they tend to take you literally. Any personal opinions or experiences should always be expressed as possibilities vs. absolutes.

The second rule of technical support is be careful of committing the post hoc ergo propter hoc fallacy (logic):

Logic 101: Post Hoc fallacy - The logical fallacy of believing that temporal succession implies a causal relation.

Cause-and-effect always requires specific validation via testing, testing again, and testing again controlling all variables until you find the specific cause. Otherwise, your conclusions may not be valid, and in most cases they're not.

The SilentAdapt feature (PelAudio Acoustic Scale Score) is an acoustic process not dissimilar to general training, but is done on-the-fly as you dictate. Nevertheless, it only applies to the Acoustic Model.

The Acoustic and Language Model Optimizer eliminates redundancies, consolidates and adapts corrections, updates both the Acoustic Model and the Language Model when it has access to and utilizes the dra files that are stored in the voice_container\drafiles folder because the dra files contain both the audio and the text associated with that audio. In essence, the Acoustic and Language Model Optimizer is a more comprehensive combination of running general training and the analysis of your writing style.

While certain aspects of these two functions overlap one another to a certain degree, the Acoustic and Language Model Optimizer is more comprehensive. Besides, I've had clients argue that they don't want to save their user profile because they're afraid that it will become corrupted. What if I were to tell them that that's OK? That would be tantamount to pulling the pin on a grenade and handing it to them. If you don't save your user profile, your user profile never gets updated, and thus never improves at all period.

If one of your clients chooses to run the Acoustic and Language Model Optimizer and it appears to be causing some problems, then that's the time to go through a detailed analysis of their hardware (especially and including the microphone/soundcard) and their system configuration. In most cases I think you will find that the problem is not with the Acoustic and Language Model Optimizer but is with the user's particular hardware/system configuration and/or the way that they are managing Dragon.

I have no problem with that. You are absolutely free to do whatever you wish. However, just remember that once you pull the pin on a grenade, Mr. grenade is no longer your friend. My way of thinking is that if you want to help your clients (truly), the best approach is to spend the time and find the true cause of the problem vs. speculation, as well as explain possible caveats, but let them learn from their own experiences without restricting them. In other words, we should never be telling clients what to do. We should be telling them what to try and then letting them make their own decisions as to whether or not to take a particular approach.

Chuck Runquist
Technical Project Manager
VoiceTeach LLC
Home of VoicePower®: We don't make Dragon NaturallySpeaking, we make it better!

"Tell me and I forget. Teach me and I remember. Involve me and I learn." - Benjamin Franklin

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

Second attempt. Please excuse if this post appears twice. Crashed on upload.

Chuck,

I have been reviewing the Dragon Help files with regard to setting up roaming profiles etc. as you advised. I had read some of them before but there is an awful lot of information to wade through and it is very easy to suffer an overload!

This comes from the section on how Dragon synchronises Master and Local Roaming Profiles. Firstly, it states that

Further down, in support of this, is a list of changes which take place upon syncronisation which includes

 

which is what you expect given that they are for use by the Optimizer. OK so far and seems straight forward however, our dra files still haven't synchronised (we haven't checked the 'Conserve archive size on network' box which would prevent their synchronisation if I've understaood that one). By synchronised, I am expecting the dra files in the local cache to appear in the equivalent drafiles folder of the master roaming profile or is that where I am going wrong? Further down  this statement apears when explaining when individual files are copied or updated to the Master Roaming User Profile or to the local cache.. 

DRA files, aco.ini; drafiles.ini

Copied to session folder if the master voice_container has space. Files are deleted after being copied; aco.ini and drafiles.ini are recreated at zero-length

 

Can you elaborate on what is meant by 'session folder'? Currently this statement may as well be gobbledygook to me.

I'm also not clear on what the acoustic data archive is as this doesn't seem to be mentioned specifically just how size of it is managed. Is it a folder or a file in a users Dragon profile? The Help files say for roaming profiles that Dragon archives corrections locally and then synchronizes the corrections to a network archive. Again, what exactly is the network archive?

 

I'm currently puzzling over the Save Acoustic Information and Always copy acoustic information to Network help files. If you tick the former should you then automatically tick the latter and what are the consequences if you don't?

 

I'm not stupid (if it was that easy we wouldn't need experts like you) , but are dra files considered acoustic information or is just corrections in this case?

Overload, Overload! Vocdelta.dat will have to wait for another day. Given that we are considering not using the Optimizer ( not because anyone has said not to use it by the way) should this be ticked as a must? Our users could be using any one of a number of pcs and so need their master profile as up to date as possible which is the whole point of roaming profiles is it not.

Slayer,

When you find the voice_container folder on the server, you will find that it contains sessions folders which in turn contain dra files from the respective dictation sessions. In effect, the voice_container folder is the archive. Archiving local dictation files involves copying them to the respective sessions folders.

Saving acoustic data is recommended, always saving them to the network is optional. If you uncheck this box, acoustic information will still be copied to the network but not at regular intervals, which, I think, is what you experience right now. All these settings involve a bit of fiddling around to find what works best in your environment and what you actually want. I had good results with ticking boxes No. 3-7 and limiting network archive size, but, as discussed above, I don't usually optimize my profiles.

Finally, the vocdelta.dat which you didn't want to ask about is actually the crucial bit since it contains the information about new or modified words in Dragon's vocabulary. I suggest ticking the "...when vocdelta.dat is full..." box to make sure that these changes are incorporated into the roaming user file regularly. If you want your users to keep up to date over several workplaces, check this box.

Stephan

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

www.egs-vertrieb.de - Speech Recognition Blog - Forum: www.immer-eine-Nuance-besser.de

Hi all,

 

I am not using roaming profiles but what I find is that  the *.dra files are being saved with my documents. I do have DNS set to save this information. Is there an easy way to delete these files once I have run the language and acoostic model optimization?

 

Pranav

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

Pranav,

Based on your post, I think you're talking about two different aspects regarding dra files.

First, in the Dragon Options dialog | Data tab if you have the "Save recorded dictation with document" option set to Always or Asked me, then every time that you dictate into Microsoft Word, Corel WordPerfect, or DragonPad, DNS saves a dra file with that filename (e.g., filename.dra) in the same location where you have save the text document of the same filename. These dra files are not used by the Acoustic and Language Model Optimizer. Their purpose is so that you can listen to your playback while editing such files.

Second, unless you have the option in the Data tab set to "Conserve disk space required by user profile (for portability)" (i.e., enabled with a checkmark) Dragon stores all of your dictation sessions as dra files in the folder in your user profile under current\voice_container\drafiles. These dra files are used by the Acoustic and Language Model Optimizer when it is run and after which converts these to archive files (*.dft & *.nwv). In this case, after running the Acoustic and Language Model Optimizer, you can, if you wish, delete these files. However, aside from cleaning up this folder to recover disk space, deleting these doesn't offer you any distinct advantage.

As regards the dra files described in the first paragraph above, you have to go to the location where you saved these files and delete the dra file associated with specific documents. However, if you have no use for these files (i.e., you don't use them for editing your documents by listening to your playback or correction purposes), then your best bet for eliminating these files is to set the "Save recorded dictation with document" option to "Never". Once you do that, Dragon will no longer save these dra files when you save your documents.

Chuck Runquist
Technical Project Manager
VoiceTeach LLC
Home of VoicePower®: We don't make Dragon NaturallySpeaking, We make it better!

"Aiming for the moon and missing it is better than aiming for the ditch and hitting it."   - Author Unknown

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

Chuck, that sums it all up rather nicely.  Can I ask you something based on a previous post you made earlier in this topic. It was, if we don't run the Acoustic and Language Model Optimizer, how do we keep the profile size as stored in RAM as small as possible? We have other applications that are RAM hungry and in some cases are having to close Dragon when running low.

The dra files that are used by the Acoustic and Language Model Optimizer are recordings that are stored on your hard drive. They are not loaded into RAM and we know of no way to change the size of your NaturallySpeaking user profile and suspect this wouldn't be a good idea if it were possible.

PS: We recommend deleting all of those uncompressed dra recordings that you saved (linked) to your documents because they waste a lot of hard drive space.

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

Click KB 2012 REV D to Download a 30 Day Evaluation of KnowBrainer 2012 

 

 

 

Hi Chuck,

Many thanks for the clarification.

<snip As regards the dra files described in the first paragraph above, you have to go to the location where you saved these files and delete the dra file associated with specific documents. However, if you have no use for these files (i.e., you don't use them for editing your documents by listening to your playback or correction purposes), then your best bet for eliminating these files is to set the "Save recorded dictation with document" option to "Never". Once you do that, Dragon will no longer save these dra files when you save your documents.

PL] Hm well I do need to listen to playback and yes, I do correct my documents in subsequent sessions so those files will stay for some time. The trouble is that I at times edit by hand and that renders the files useless since Dragon loses its association with the document or so I think.

 

Pranav

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

Pranav,

correction by hand (not using the DNS correction features) will inevitably destroy the sound files since it results in text with no recording attached to it. You can minimize the risk by going into Options - Data - Advanced and check "Alweays preserve wava data". Be aware that this may result in other behaviour that you may find strange. For instance, this setting will preserve the recording even when the associated text has been deleted.

Stephan

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

www.egs-vertrieb.de - Speech Recognition Blog - Forum: www.immer-eine-Nuance-besser.de

Stephan,

no matter how hard you try to convince us, you don't have the faintest idea of how speech recognition works, and in fact you keep confusing everything you are picking up along your way through it. As regards the option you have referred to in your previous post, "always preserve WAFV data" for that matter, here is how this is described per the SDK (but also please don't confuse your alter ego "dsk" with it):

Regards,
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!

Rüdiger,

as opposed to you, I have installed enough dictation workflow systems to know how the "Always preserve wave data" option works, and why it may be helpful in some situations.   Moreover, "Always preserve wave data" is different from the "Record wave data between utterances" option in the Administrative settings which you seem to be referring to. If you wish, I can send you a dra file to show how it works.

So go play with your SDK and stop insulting me.

Stephan

PS  For the benefit of others: checking the "Always preserve Wave data" option preserves all dictation, even if portions have been dictated over, manually edited or deleted. If you have issues with deferred correction, this option may help, but of course you'll have to listen all over again to the bits that were already edited out. 

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

www.egs-vertrieb.de - Speech Recognition Blog - Forum: www.immer-eine-Nuance-besser.de

Stephan,

sorry, you are correct and I was wrong. My apologies for insulting you.

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!

Chuck,

If only it did. Last time I ran it, the archive size was actually larger than before. What you can do is manually delete all the files that the Optimizer ran on.

Stephan

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

www.egs-vertrieb.de - Speech Recognition Blog - Forum: www.immer-eine-Nuance-besser.de

Stephan,

True enough depending upon how you're looking at your user profile. Every time you run Accuracy Tuning (Acoustic and Language Model Optimizer) the overall size of your user profile stored on the hard drive will increase by a degree because all of this information is consolidated into your stored profile, including the dft and nwv files, which contain the archives of your dra files after running the Acoustic and Language Model Optimizer.

Yes you can delete these as long as you have saved your user profile after running the Acoustic and Language Model Optimizer. If you keep them, then they will be updated and use the next time you run the Acoustic and Language Model Optimizer. This will reduce the overall size of your user profile stored on the hard drive. However, but I was referring to is the fact that your active user profile, what it is running in RAM when you load Dragon and open your user profile, does get consolidated and reduced in size.

In my case, he for I ran the Acoustic and Language Model Optimizer my user profile occupied 1.3 GB of RAM. After running the Acoustic and Language Model Optimizer the amount of RAM used by Dragon for my user profile was reduced to 930 MB. Regardless, unless you elect not to create and store dra files via the option noted [i.e., Conserve disk space required by user profile (for portability)], your user profile is still going to increase in size gradually over time if only by virtue of the fact that making corrections, adding custom words, adding custom commands, running additional general training, etc. will gradually increase it, even though to a lesser extent.

It basically boils down to the resources the user has to work with (i.e., RAM and disk storage space) and what they wanted do as far as controlling the size of the active user profile. The other point that's important to understand is that it depends upon exactly to what slayer is referring when he says that user profiles are getting >1 GB. Is he referring to the amount of RAM that Dragon is using, or the size of user profile on the hard drive. When I originally created my current user profile it was 384 MB. Before I ran the Acoustic and Language Model Optimizer (Accuracy Tuning) the last time it was 1.45 GB. After running the Acoustic and Language Model Optimizer, saving my user profile, and deleting all of the dft and nwv files in the drafiles folder, my stored user profile was 834 MB. So, over time, regardless of how you handle the dra files, your user profile is going to increase in size. That is simply a given. However, the dra files stored in the drafiles folder does dramatically increase the size of your user profile stored on your hard drive.

It's still all depends on how you want to manage your user profile. Regardless, the size of your stored user profile is not going to remain static over time. The size of your stored user profile increases dynamically depending upon how long you use it.

Chuck Runquist
Technical Project Manager
VoiceTeach LLC
Home of VoicePower®: We don't make Dragon NaturallySpeaking, We make it better!

"Better a diamond with a flaw than a pebble without one." -  Chinese Proverb

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

Chuck,

you're absolutely correct. Thank you for clarifying.

Stephan

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

www.egs-vertrieb.de - Speech Recognition Blog - Forum: www.immer-eine-Nuance-besser.de

 

In answer to your quote, she was referring to the size of the user profile on the hard drive.

This discussion has been very helpful for me, because I have a 64Gb SSD drive, so I have to do everything I can to conserve disk space. Deleting the .dra and .nwv files has allowed me to claim back quite a large chunk of disk space.

Thanks!

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

DNS 12 Premium
VXi Tria V-DC w/ VXi Full Duplex USB Pod
Windows XP Pro SP3

Bodhi,

Nuance guards the DRA file format jealously, so there is no way of opening these files short of finding someone with a Dragon Professional. Ask you local dealer.

Still, there is no guarantee that the dra files you've found actually contain your work, so be prepared to do it all again... and then save your work at regular intervals. You can say something like "save file", too.

Stephan

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

www.egs-vertrieb.de - Speech Recognition Blog - Forum: www.immer-eine-Nuance-besser.de

                        Welcome to the World’s Most Popular Speech Recognition Forum

Trusting Windows or NaturallySpeaking to keep running while spending hours on a project is equivalent to riding a motorcycle naked through the heart of New York City during rush hour. Common sense dictates that you should save your work every 10 minutes or so. Our 1st recommendation would be to use Microsoft Word or WordPerfect rather than DragonPad because professional word processors contain more features including a default setting which automatically saves your document every 10 minutes. This would give you 2 backup plans.

Not even DNS Pro can open dra files which are strictly proprietary wav file recordings. If DragonPad crashes but NaturallySpeaking manages to stay open, you could view your dictation by saying view recognition history but even if you're lucky, the Recognition History box only temporarily saves your last 1000 dictations, 1 phrase at a time. Enfortunately, you are most likely SOOL. Sorry about that…

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

Click KB 2012 REV D to Download a 30 Day Evaluation of KnowBrainer 2012