Dragon Speak to Continue Using This Profile You Must Restore a Previous Version
| | |
| | This originally appeared on my blog https://trainyourdragonblog.wordpress.com/2017/03/02/repairing-a-broken-dragon-naturally-speaking-profile/. Repairing a Broken Dragon Naturally Speaking ProfileMy Dragon profile self-destructed again. I'm pretty sure it is related to adding numerous dictation sources over the time I have been using the product. I just completed a rather unsatisfying call with technical support from Nuance. Their position is that the profile was corrupted through no fault of their own. I.e. they think it is my fault or Windows fault. The reality is Dragon is corrupting its own profiles, they should be able to save their profile without fault whether it windows crashes or not (which has not been the case). Further, he said there was no way to rescue my vocabulary customizations from the last several years of Dragon use, nor my custom commands. His solution was for me to create a new profile and start over. So if you find yourself using Dragon naturally speaking, and you find all of a sudden you cannot turn the microphone on, or manage dictation sources, the following might help you out:
That should allow you to have a new profile with your old commands and your old vocabula ------------------------- Doug Ransom |
| Join to Reply : Top : Bottom | |
| | |
| | There is one small problem with your procedure. It's not possible to export your personal vocabulary if you cannot open your user profile; ala catch 22. Did you know that by default, every 5 times you save your user profile, Dragon overwrites your backup profile which means there is an 80% chance that your backup profile is intact and you don't have to create a new user profile or export and import your personal commands and vocabulary. You'll find 2 options in the 1. Begin by opening Windows Explorer/File Explorer (depending on your OS) You have an 80% chance of being able to open your profile. PS: You will find a step-by-step on removing out of date Dictation Sources when you search for "dictation source" in the previously mentioned Quick Tips. ------------------------- Change "No" to "Know" w/KnowBrainer 2020 |
| Join to Reply : Top : Bottom | |
| | |
| | +1 Lunis. I don't think that your user profile auto-destructs because of the number of dictation sources. In my experience, user profiles go west because of issues while saving the profile, or due to illegal entries in the vocabulary. If a source were the culprit, you'd still be able to use the other sources. I'm confronted with defective user profiles almost on a daily basis - it's part of my job description. Fortunately, most professional users in Germany have Roaming profiles (which seem to be completely outmoded in the US), so restoring them is usually a matter of renaming a folder and copying that folder to the network Roaming location. In the rare case that this doesn't work, I check for the following: - any irregular files and folders, e.g. ending in "_dgn_renamed" - rename those back to the correct name The bottom line is that if you don't want to pay for Dragon Professional Group, you better get a safety policy in place for your Dragon profiles. Stephan ------------------------- www.egs-vertrieb.de - www.spracherkennungscloud.de |
| Join to Reply : Top : Bottom | |
| | |
| | I did find that the profile I had, and a copy to a second computer, had problems after I added what seemed to be one too many audio sources. I was able to repeat this on both computers as I had bought a new microphone and decided to export my profile from my office to home computer around the same time. I first ran into the problem at home, adding a new microphone seemed to make the profile fail. Then I ran onto the same problem on my work computer. A previous commenter was also correct in that if you do not have a backup of your profile, with your vocabulary and commands intact, you may have trouble even exporting your vocabulary and commands. As Dragon puts a lot of this information in a hidden folder, it may not be in your regular backups depending on the software you use for your daily backups. ------------------------- Doug Ransom |
| Join to Reply : Top : Bottom | |
| | |
| | Could it be that the usage of the Auto Transcribe program can screw things up? As I experience it, Auto Transcribe uses another dictation source than when I am dictating at the computer directly. And it dos not work while Dragon loaded. |
| Join to Reply : Top : Bottom | |
| | |
| | It should be OK to run the AutoTranscribe Folder Agent and a Dragon user profile at the same time but Dragon automatically disables the AutoTranscribe Folder Agent so that you can dictate but not transcribe; other than by going to through the Transcribe Recording menu. When you wish to transcribe through the AutoTranscribe Folder Agent, you need to close Dragon 1st. Think of it is starting your car by removing the key from the ignition. Once the AutoTranscribe folder finishes, you should be able to relaunch Dragon and your user profile. You cannot use the AutoTranscribe folder and a Dragon profile simultaneously. That would be equivalent to attempting to dictate with 2 different user profiles simultaneously. If after you reboot, your user profile exhibits problems, you always have the option of switching to your backup profile. You'll find a step-by-step in the Quick Tips when you look up "switching". ------------------------- Change "No" to "Know" w/KnowBrainer 2020 |
| Join to Reply : Top : Bottom | |
| | |
| Alan Cantor | I gave up on repairing misbehaving profiles. Here is my quick way of dealing with a problematic profile. (It may not be the best approach if you change a lot word properties; but the procedure is fine for me because the only word I change the properties of is "new line.") 1. Develop the habit of exporting your custom words and phrases. How often? It depends on how frequently you add words and phrases. For example, I add five or 10 words per month, so I export my words and phrases two to four times per year. If you are constantly adding words, increase the frequency. 2. Maintain a plain-text writing sample. Mine is about 50,000 words long. I add a few dozen paragraphs three or four times a year. When a profile goes south, create a new profile, import your custom words/phrases, let Dragon analyze your writing sample, make any setting changes you want to make via Tools > Options and Tools > Autoformatting Options, exit Dragon, copy over custom commands (as you describe), restart Dragon, and voilà ! An accurate profile in five or ten minutes! |
| Join to Reply : Top : Bottom | |
| | |
| | Hi May i ask, how many corrupted dragon 15 profiles are you getting, is this a regular occurance. Thanks |
| Join to Reply : Top : Bottom | |
| | |
| Alan Cantor | I don't find my Dragon 15 profiles degrade any quicker, or are any less fragile, than in earlier versions. I estimate that in the normal course of things, I create a new Dragon profile every six months to a year. As I mentioned earlier in this thread -- was it really three years ago?? -- the process is fast. My preference is to spend five minutes creating a new profile than to spend five minutes trying, usually in vain, to resurrect a misbehaving profile. |
| Join to Reply : Top : Bottom | |
| | |
| | Same here, if that often. Your profile really shouldn't be degrading quickly. If it is, my first suspicion would be that your microphone or soundcard are not up to par. If they are adequate, then I would worry about your dictation practices. |
| Join to Reply : Top : Bottom | |
| | |
| | with DPG 15.6 I find that I need to rebuild my profile every one to four weeks. If I'm just using Dragon/KnowBrainer/MyCommands, the longer period of time. But I find when I start editing or creating commands in KnowBrainer, the profile degrades. That might be coincidence. When I rebuild my profile (1) change all the options to the settings I want - This is quite important, various options like Disabling HTML commands and setting disk space or play back to zero seem to greatly improve reliability (2) I add custom words - I follow Lunis's and Leslie's advice, trying to make written\\spoken-forms that don't require training. mostly it works, but not always. I've been keeping a log of what vocabulary terms needed training - for the most part my custom words live in a TXT file, and are hence easy to version control and diff - some custom words have special properties likebefore after. For those I save the custom words XML file (3) and then delete the standard vocabulary terms that get in my way. e.g. Cindy\\DELETED conflicts SIMD". I keep custom word forms such as that before to remind me which standard vocabulary terms need to be deleted. My BKM is to load the fresh(ish) profile, save the custiom words TXT, git diff, manually add the changes, grep the DELETED forms, delete those. or a completely new profile this takes about an hour I have been trying to create a nice set of saved/exported profiles that are a completely new profile plus the minimum set of these changes. E.g. after restoring from BaseLineV1, I make the changes above, save that as BaseLineV2, and then work. When the new profile branch inevitably corrupts, I repeat the process starting from BaseLineV2. I just did that today. It took 26 minutes - Although perhaps 10 minutes of that were spent doing things like diff'ing custom.ini, and trying to save any vocabulary in the profile I suspected of corruption. It turned out that there were no changes, but it took a long time because the biggest symptom of possible corruption in my profiles is huge delays, 1 to 2 minutes, sometimes more. -- MC: first suspicion would be that your microphone or soundcard are not up to par. I have a damn good microphone, and I have submitted to the tests that y'all recommend. That does not seem to be the problem. MC: If they are adequate, then I would worry about your dictation practices. if you can tell me what dictation practices I should not be practicing, I would like to practice not practicing them. ------------------------- DPG15.6 (also DPI 15.3) + KB, Sennheiser MB Pro 1 UC ML, BTD 800 dongle, Windows 10 Pro, MS Surface Book 3, Intel Core i7-1065G7 CPU @ 1.3/1.5GHz (4 cores, 8 logical, GPU=NVIDIA Quadro RTX 3000 with Max-Q Design. |
| Join to Reply : Top : Bottom | |
| | |
| | Sounds like your microphone is okay, but I'm no expert on microphones. Maybe others here know this microphone and can offer their thoughts. I know that some microphones that are high quality for some purposes, such as podcasting, don't work for speech recognition. On the dictation practices, are you turning your microphone off when you are not dictating? Some people have theorized that hearing extraneous noises can cause problems with your profile. On the creating a new profile front, maybe you should export a profile as soon as you have it set up and also export a version of your profile when it is working well, so that you have backups and don't have to waste 26 minutes removing vocabulary entries and fixing *.ini settings. At the very least, I would export a pristine version of your vocabulary as a backup. |
| Join to Reply : Top : Bottom | |
| | |
| | We have only experienced one corrupted user profile (minor or major) in 2 decades so we are beginning to think that we might be doing something right. We suspect the most common Dragon corruption source is allowing Dragon to save your user profile when closing. You used to be able to say yes or no but now it's automatic. In theory, nothing should go wrong because Dragon will not exit until your user profile has been saved. We suspect this procedure. You never need to save your user profile when you create a command but if you make a change to your vocabulary or train, you might consider saving your user profile at that moment by sayingSave User. We pretty much never close Dragon or KnowBrainer but if you feel the need, say Close Dragon and KnowBrainer. This command will do exactly what you think but not save your user profile. More often than not, we don't save our Dragon profile but we find this procedure works well for us. ------------------------- Change "No" to "Know" w/KnowBrainer 2020 |
| Join to Reply : Top : Bottom | |
| | |
| | We paid millions in taxes, but the IRS didn't treat us kindly. We have also done an excellent job on handling Covid 19, better than any other nation. Windmills cause cancer. Raking the forests will prevent wildfires. When an 800-year-old cathedral is on fire, water bomb it. Wearing masks is for losers and suckers only. Our polls are at an all-time high, and we will win bigly. -------------------------  No need to buy if all you want to do is try ... DragonCapture KB Download (Latest) |
| Join to Reply : Top : Bottom | |
| | |
| | Lunis provides much valuable advice, and in the above is something that I'm paying much more attention to. I cannot agree that these problems occur only at the moment we are saving a profile to disk. I can be going along quite happily for quite some time, but then things start going bad. Using whatever state is in memory, not saved to disk explicitly, not saving to disk at all unless Dragon is doing it behind my back. however, if the profile is corrupt in memory, then not saving it should result in going back the version on disk that might not have that corruption Not saving a profile that seems to be having problems is the simplest and cheapest backup. It doesn't always go back far enough, but it's often good enough to give yourself a few more days. What about losing good stuff if you don't save your user profile? See below. ---+ How to save good stuff without saving the user profile saving a user profile does good things like saving custom words that are being added, options that have been changed, and training. But one of Lunis's other good pieces of advice is to avoid training and that is standing me in good stead as far as it works. When I add custom words with a simple written/spoken forms, I try to remember to export them as a TXT file as quickly as possible. I have not fully automated that, because I don't want to overwrite already saved stuff, and I'm not quite sure that I want to check things into git immediately without looking them over. Sometimes I have added custom words by mistake. And although git is great, it's still a pain to have to go through your history to delete crap. When I add custom words that have special properties that cannot be represented by the custom word TXT file, I do the same thing through a custom word XML file. I would always use XML, except that git does not yet understand UTF-16 properly. when I delete standard words, I add "fubar\\DELETED conflicts foobar" terms as custom words, so at least I can fairly quickly find the words and re-delete them. So the only things that I am missing by not doing "save user profile" are [A] not saving training, which as Lunis and Leslie have taught is actually a good thing, options, and [C] deletions of standard vocabulary terms. I do have to reapply the deletions of standard vocabulary terms. But as mentioned elsewhere I do that incrementally, trying to track a baseline that has all of my deletions. also, once I have applied my deletions from a working profile that may or may not be partially corrupt, to a known good fresh profile, I can always restore from that profile without having to reapply anymore deletions. And I can always detect which deletions are needed because of my habit of putting "fubar\\DELETED conflicts foobar" entries in my custom vocabulary. Also from time to time, I export a completely new vocabulary, supposedly independent of the profile. although I haven't quite perceived any big advantage between separating vocabulary from profile and just trying to save a known good profile. So that leaves the main lossage, apart from training, as being options that are in the profile. And I don't change options very much. I also now have the habit of diffing custom.ini ------------------------- DPG15.6 (also DPI 15.3) + KB, Sennheiser MB Pro 1 UC ML, BTD 800 dongle, Windows 10 Pro, MS Surface Book 3, Intel Core i7-1065G7 CPU @ 1.3/1.5GHz (4 cores, 8 logical, GPU=NVIDIA Quadro RTX 3000 with Max-Q Design. |
| Join to Reply : Top : Bottom | |
| | |
| | For the most part, I turn my microphone off when not dictating. e.g. I have interfaced a "Press-To-Talk" button from my wrist. Works nicely, battery life sucks when wrist-worn e.g. I fairly reliably turn the Dragon Mike off, again by pressing a button on my wrist. No battery life problem there, momentary. e.g. I nearly always hang my Sennheiser headset up on its charging stand when I step away - obviously, I want it to be charged up when I need it. BTW, I wonder if there is a difference between "Dragon mic off" and "Dragon mic on, but physical mic off" - if in the latter case crap is still getting sent to Dragon. -- Apart from such habits, I plan soon to code up an automatic timeout for the Dragon mic. the problem has been detecting idle. I have several heuristics that seem to be good enough, including looking at Dragon.log, and also keeping track of last use time and commands I have written myself. unfortunately, setting the timeout to less than five minutes can be quite annoying, and letting it keep listening for five minutes of nonuse also seems to cause problems. -- Circa 30 minutes later: just experienced an example of why I do not always turn the Dragon mic off: Dragon/KnowBrainer suddenly stall. Past experience shows that it often stalls for 1-5 minutes, and then resumes. Often I hang up my mic and go away to do something else, planning at least to export my custom words when I get back. However, life often intervenes, and I continue working on the old profile that had that bad stall. Which gets worse. I am trying to follow Lunis' advice, and almost never save my current user profile. Not always successfully. Sometimes saved unexpectedly. That's what I do. The 26 minutes was going back to the profile that I saved as soon as I had it set up most recently - merging in vocabulary changes. 10 minutes was thrashing, much of the rest of the time was renaming files to keep track. I have scripts to do this when my profile is running, but when I have a corrupted profile the scripts may not be running. plus, more and more I suspect that it is running certain types of commands that is associated with the corruption, so I am trying more and more to do all of my baseline profile maintenance without KnowBrainer running. It took me a while to realize that I need to make pretty much every vocabulary change at least twice: First to my working profile, Then to my "baseline" saved profile. Workflow: Baseline started with a new profile. I set options, added custom words, deleted standard words, And saved that baseline. I then fork off a working branch/profile. While working on a profile branch, I regularly save profile, custom words XML and/or TXT files, etc. Eventually, when working profile branch becomes unusably slow. I step back to baseline. Merge in custom words added, delete standard words, tweak options. Save/export that as the new baseline. And then I fork off a new working profile branch, and repeat. Unfortunately, even this "baseline to which all important changes have been made, but on which no active work has been done" eventually seems to get slow. So I start with a fresh new profile - and then make all of the changes to that in one fell swoop, rather than as a dozen or so different layers. This workflow is helping reliability a lot. Although ity is still quite painful. --- BTW, my two most recent step back occasions were associated with air quality/forest fire smoke events. When we crank up our air cleaners when it gets hard to breathe, shortly thereafter Dragon starts dragging. Rerunning check microphone in the new audio environment extends the life of the old profile by a few days. Not much, but it helps. Nvidia RTX Voice's noise suppression helps a lot. I suppose this is understandable, but I continue to be surprised that no active noise cancellation wireless headset is recommended for speech recognition. ------------------------- DPG15.6 (also DPI 15.3) + KB, Sennheiser MB Pro 1 UC ML, BTD 800 dongle, Windows 10 Pro, MS Surface Book 3, Intel Core i7-1065G7 CPU @ 1.3/1.5GHz (4 cores, 8 logical, GPU=NVIDIA Quadro RTX 3000 with Max-Q Design. |
| Join to Reply : Top : Bottom | |
| | |
| | For the most part, I turn my microphone off when not dictating. ... e.g. I nearly always hang my Sennheiser headset up on its charging stand when I step away - obviously, I want it to be charged up when I need it. ... BTW, I wonder if there is a difference between "Dragon mic off" and "Dragon mic on, but physical mic off" - if in the latter case crap is still getting sent to Dragon. I can only hope that hanging up my headset mic on its charger stand is asm good as turning the Dragon mic off. Sometimes, oftenI cannot turn the Dragon mic off in a timely manner, but I can hang up the headset (or mute it). E.g. just right now, I created a new KnowBrainer global command. It will probably take 1-5 minutes before I can turn off the DragonBar mic again. Yep... 3 minutes that time. Oftentimes I hang up my headset and walk away. ------------------------- DPG15.6 (also DPI 15.3) + KB, Sennheiser MB Pro 1 UC ML, BTD 800 dongle, Windows 10 Pro, MS Surface Book 3, Intel Core i7-1065G7 CPU @ 1.3/1.5GHz (4 cores, 8 logical, GPU=NVIDIA Quadro RTX 3000 with Max-Q Design. |
| Join to Reply : Top : Bottom | |
| | |
| | When you create a KnowBrainer global command, it may take around 16 seconds to deploy but after it has deployed for the 1st time, the 2nd deployment should be instant. It certainly shouldn't take 3 minutes. Our computer is 7 years old but also the fastest monster on the market at the time. Try deploying your command again and you'll see what we mean. Having said that, we are working on that 16 seconds to 3 minutes. We have 2 solutions that we are still testing. In the meantime, instead of creating your global commands in KnowBrainer, create them in Dragon. Ver. 15.6 has the same scripting engine as KnowBrainer but you won't have to deal with the initial delay and better yet, it will be much easier to update KnowBrainer with the KBCommandsUpdater ------------------------- Change "No" to "Know" w/KnowBrainer 2020 |
| Join to Reply : Top : Bottom | |
| | |
| | How exactly do you want it to operate? Specify the details, and I will write if for you, probably as an extension to DragonCapture, thus "clean", from a Dragon programming perspective, and at no cost for you. ------------------------- No need to buy if all you want to do is try ... DragonCapture KB Download (Latest) |
| Join to Reply : Top : Bottom | |
| | |
| | What I would like: Turn the Dragon mic off N minutes after the last non-error utterance - after the last text, or useful command. Since I do not yet have any tools that are hooked to an actual recognition, I can only monitor the Dragon.log last modified time and the last time one of my commands that I do have a hook in has been run. Come to think of it: my AHK scripts often sit in between Dragon/DragonCapture and the OS/application, so I can probably track a most recent time there. I'm not sure that I capture everything, but better something the nothing. Also, if I had a way of seeing if Dragon was in "sleep" state, I could use Dragon's existing timeout and then just promotes sleep to microphone off. although that would require a polling loop, e.g. pulling every minute or so, and such pulling loops are notoriously unreliable and power wasteful. --- if you wrote such a tool, Rudiger, I would be happy to give it a try. But I will probably still continue writing my own. I am finding it so enabling to escape from Dragon scripting into AutoHotKey, making it so much easier to add features like these, even though I still have some restrictions because I am not hooking everything. ... Here're more wish list features: Possibly pop-up a non-focus capturing DialogBox or other notification indicating when you have timed out. Also, more and more of my tools are logging such events into a log file, that I can merge with the Dragon.log file and try to figure out what's going wrong such a notification can be annoying in the case where things are going well. Perhaps more important would be to provide such notification when the timeout occurs but Dragon's microphone could not be turned off. Regularly, during these things that you tell me that I should not have, I am unable to turn the DragonBar microphone off for 5 to 10 minutes at a time. ------------------------- DPG15.6 (also DPI 15.3) + KB, Sennheiser MB Pro 1 UC ML, BTD 800 dongle, Windows 10 Pro, MS Surface Book 3, Intel Core i7-1065G7 CPU @ 1.3/1.5GHz (4 cores, 8 logical, GPU=NVIDIA Quadro RTX 3000 with Max-Q Design. |
| Join to Reply : Top : Bottom | |
| | |
| | AG, I have never been a fan of polling and looping, and avoid it like a plague. Automation is all about catching events, and acting accordingly. Fortunately, the Dragon API allows for it. DragonCapture wouldn't exist if it weren't for that, and the same is true for the semi-automatic Dragon dictation box, by the way. I am currently doing a video where this will be demonstrated also. Also, keep in mind that writing to the Dragon logfile goes on in the background, also when the file is open, and there is whole lot going into it, so you may miss things when doing it that way. I will gladly put it on my to do list. If it works as expected, maybe others will find it useful, too. ------------------------- No need to buy if all you want to do is try ... DragonCapture KB Download (Latest) |
| Join to Reply : Top : Bottom | |
| | |
| | BTW, I just "stepped back" to my most recently saved baseline. It took 12 minutes this time. It would probably have taken less time, if KnowBrainer had not been hung and had been accepting commands. it will take less time, if all of those things I have automation for in KnowBrainer are actually controlled from AutoHotKey so that they will work when KnowBrainer is not working. I did this partly just to see how fast I could do it, partly because I was seeing some annoyingly long delays. Hmmm I'm still seeing those long delays even on this fresh profile. So I will probably have to go back to an even fresher profile that will require more work to merge in changes made. but I will leave that until the weekend. ------------------------- DPG15.6 (also DPI 15.3) + KB, Sennheiser MB Pro 1 UC ML, BTD 800 dongle, Windows 10 Pro, MS Surface Book 3, Intel Core i7-1065G7 CPU @ 1.3/1.5GHz (4 cores, 8 logical, GPU=NVIDIA Quadro RTX 3000 with Max-Q Design. |
| Join to Reply : Top : Bottom | |
| | |
| | Out of the blue, I received several ridiculous Dragon error messages regarding my profiles. I successfully followed these steps (as listed above) to restore my working profile, which includes its vocabulary and custom commands. I cannot recommend strongly enough to make regular backups of all your Dragon customizations. It is garbage software, which is why this forum is so active with problems and solutions. |
| Join to Reply : Top : Bottom | |
| | |
| | It would appear that you are using DPI 15.3 because Ver. 15.61 moves the Ver. 15.3 user profile to ~ %LOCALAPPDATA%\Nuance\NS15\Users\[your username] folder ------------------------- Change "No" to "Know" w/KnowBrainer 2020 |
| Join to Reply : Top : Bottom | |
03/03/2017 10:12 AM 
.jpg)
Source: https://knowbrainer.com/forums/forum/messageview.cfm?catid=4&threadid=23976
0 Response to "Dragon Speak to Continue Using This Profile You Must Restore a Previous Version"
Post a Comment