- Messages sorted by: [ date ][ thread ][ subject ][ author ]
- Next message: Holland St.John: "Re: Training new commands in NaturallySpeaking"
- Previous message: Eric S. Johansson: "Re: Speech recognition on Linux"
- In reply to: L.C.Jackson@@psd.maff.gov.uk: "Headset tip (was Mic Conundrum, New To List)"
- Next in thread: Holland St.John: "Re: Training new commands in NaturallySpeaking"
This is a problem for words that sound alike but spelled differently. I
presume that when you select "check male" and look in the correction box
that "check mail" is first on the list.
The surest, quickest way of solving this is to use the topic editor
(vocabulary editor) to select "male" and delete that word (as I am a man
I have somewhat of an aversion to recommending a woman delete this
word... but that doesn't have anything to do with Nat). Of course, this
is not a very good solution if you use "male" in your dictation.
Option 2: Use the topic editor to define a custom phrase "check mail."
This should work without the problems of the initial option. Recognition
should be good, but not as good as option 1.
Option 3: Generate a new document with the phrase "check mail" written
about 100 times. Then use the Topic Builder to rebuild your language
model, being sure to include all your old documents in addition to the
new one. This will make a new language model with "mail" being given
preference over "male", and with "mail" being associated statistically
with "check".
Let me know how it works out.
Holland
Dr Jenny Woods, ERU, RAL wrote:
>
> I'm sure this is a fairly basic question, but how do
> I train my new commands in NaturallySpeaking?
>
> For example, I wrote a little macro which typed
> {ctrl+m} when I said "check mail " in Eudora. It worked
> fine at first, but now I am getting the phrase recognised
> as "check male " (!). In the help file it tells you to
> use the train words on the tools menu. But this doesn't
> seem to be working appropriately for new commands. I
> have also tried holding the control key down, but all
> I get is a Question mark as my response.
>
> Thanks, Jenny
> -------------------------------------------------
> Dr Jenny Woods
> Energy Research Unit
> Rutherford Appleton Laboratory
> Chilton, Didcot, OXON, OX11 0QX
- Next message: Holland St.John: "Re: Training new commands in NaturallySpeaking"
- Previous message: Eric S. Johansson: "Re: Speech recognition on Linux"
- In reply to: L.C.Jackson@@psd.maff.gov.uk: "Headset tip (was Mic Conundrum, New To List)"
- Next in thread: Holland St.John: "Re: Training new commands in NaturallySpeaking"
 |  |