Registering and logging in removes this ad.
Registering and logging in removes this ad.
DNS 9's new "Language Model Optimizer"
Submitted by Mephistopheles on Sun, 10/08/2006 - 16:02.
I just discovered this new feature in DNS 9 preferred. The online help describes it as follows:
"Perform Language Model Optimization updates your language model. The language model contains statistical information that predicts which words are most likely to occur in the context of the user’s speech. Language Model Optimization uses text extracted from a user's .DRA files to add commonly used word sequences to the language model"
I guess the idea here is similar to the old "vocabulary builder", but that instead of using word processing or e-mail files you have already dictated, you use your actual more recent dictation to determine word and word pattern frequencies. Not sure if there is necessarily a big advantage to this, but I guess it would be a convenient way to keep DNS up to date on word frequencies and avoid the need to rerun the Vocabulary Builder (or similar tool) with more recent documents.
By the way, I'm not sure what options you need to have selected to make this a worthwhile process. Is DNS automatically saving the dra files if I select store corrections an archive? Or do I need to check something else?
Meph
First, I am a confirmed
First, I am a confirmed solipsist, so I will not apologize for replying to my own post.
Second, after spending some more time with the online help, it looks as though the acoustic optimizer just has a more accurate name, but performs essentially the same functions as in DNS 8. Both acoustic and language optimization work by processing correction data and general training, so it does not appear that your actual recent dictation is processed other than those two items. Running it is supposed to be superior to making corrections individually because of the volume of data, and because you can reject the changes if they do not improve accuracy.
Meph
In any event, for some of
In any event, for some of us, the Language Model Optimizer part doesn't work. At least for me, it stopped working a month ago, gave me a message that "%1 not found", and won't modify my files. Apparently some file or data gets corrupted, and the Language Model Optimizer won't work. Doesn't seem to matter, because my accuracy is quite high.
The Acoustic Optimizer works just fine.