- Messages sorted by: [ date ][ thread ][ subject ][ author ]
- Next message: Judy Evans: "Re: Is it just me?? and apology"
- Previous message: Lynda-L: "Correct That Command revisited"
- Next in thread: gNAt: "Re: Is it just me??"
Date: 19/7/98
To: Voice-users Group
Fr: Charles Horth
Email: Charles_Horth_Biomed@@compuserve.com
Re: Do dragons live in swamps?
Hello NatSpk users!
Holland,
We should really appreciate your pioneering work while we wait
apprehensively.....
Holland St.John <holland@@rattlers.net> wrote:
>It could be worse, I spent many hours training what I assumed was an
upgraded "BestMatch" and puzzled why I wasn't seeing the improvements that
were supposed to be there but weren't. I would have saved considerable time
and effort If the good Dragon had told me during installation of V3 the
version 2 files it so happily "upgraded" were just upgraded to run with
version 3, not upgraded to BestMatch.<
Judy Evans <jae2@@york.ac.uk> wrote:
>I look forward to many such "first encounter" stories... :)
-- So (or is this a separate point?) you don't know how much improved
Dictate and Command separation are? -- I have no real problems with them
now, but a lot of people seem to. <
Holland St.John <holland@@rattlers.net> wrote:
>I had no problems either with command/dictate separation with v 2.
Actually I have a bit of a problem with ver 3 wanting to display "select
word word word" when I want it to select a phrase. Somewhere I read this
was a side effect of more flexibility in the commands. But I am not sure it
is not just a teething problem.<
Judy Evans <jae2@@york.ac.uk> wrote:
>I didn't think you had -- but I wondered if you'd noticed anything about
Command and Dictate, to encourage people who have the problems.
Does the Select problem apply to the new "paragraph and a bit" feature? --
the Select from X to Y? -- or is it the old Select "the white elephant's
house" business?<
>-- BTW what are your hardware specs?<
Holland St.John <holland@@rattlers.net> wrote:
>I have a PII, 233, 128ram, TB Tropez, TalkMic. Thought it was fast until
ver 3 came along. It is still ok, but I now wish I had a PII 400.<
Judy Evans <jae2@@york.ac.uk> wrote:
>Oh dear. Yes, I did know that configuration; I thought perhaps you'd
managed to upgrade to a PII400; sorry.
- -- I got a PII 266 128mb ram, about 2 weeks before the swine, sorry
Dragon Systems, announced Nat. 3.0.....
I shall be slightly ahead of you there as by the time 3, and esp its
upgrades, get over here, 400s will be cheap...
----------------
Holland St.John <holland@@rattlers.net> wrote:
>It could be worse, I spent many hours training what I assumed was an
upgraded "BestMatch" and puzzled why I wasn't seeing the improvements that
were supposed to be there but weren't. I would have saved considerable time
and effort If the good Dragon had told me during installation of V3 the
version 2 files it so happily "upgraded" were just upgraded to run with
version 3, not upgraded to BestMatch.<
Lynda-L <lyndal@@airmail.net> wrote:
>Wow, that would be pretty frustrating. So, I'll bite--what do you do when
upgrading from version 2.02 to version 3 to ensure that Best Match is
incorporated and running? It can't be very obvious or you wouldn't have
missed it. Also, how do you check afterward so that you know it is
actually being utilized?
I would like to be sure about this, because I am sick to death of training
voice files (after reconfigured installations, after new mikes, new sound
cards, new computers, etc.). Like Rose stated a long time ago, I pretty
much know all the training texts by heart. I've finally got what I like to
think is a good mix of computer/mic/soundcard. Therefore, with version 3.0
(when it arrives!), I would really like to get the training right the first
time around. <
Holland St.John <holland@@rattlers.net> wrote:
>That's the irritating part. There is no way to check if a particular set
of speech files is BestMatched or not that I can find. But, it is never
true that your "upgraded" ver 2 files are BestMatched. The best thing to do
is to save your list of custom words and phrases from your ver 2
vocabulary, perform the upgrade to ver 3, and when finished delete all the
users and topics that show up in the version 3 menus. Then start a new user
and new topic, picking the BestMatch options for each of them. Do the
General Training for your new user/topic, add back your custom
words/phrases/voice shortcuts, do vocabulary building and get going on
tuning and training your new voice files.<
In my efforts to get claification on upgrading to ver 3.0, so that I could
get US spelling, I got the impression that > when you upgrade, all the new
stuff is installed over the top of the old files - except in the case of
the speech files and vocabularies.
They are handled differently - because they are such a specific type of
file - so that they can pick up any new pieces but at the same time
maintain the information they already had about the user's voice etc. One
could completely uninstall the UK 2.02 version, including the speech files,
and then install the US 3.0 - but that would be basically like performing a
completely clean installation and one would have to start from scratch with
training and adaptation etc<.... but I would get the spellings I wanted.
However.............
On July 15, "Patrick Henebry" <tphenebry@@csi.com> wrote:
>Note: since Dragon Systems has changed the language models in release 3.0
of all of its NaturallySpeaking products as part of its new 'BestMatch'
technology you would need to retrain NaturallySpeaking from scratch once
the 3.0 release is installed. Therefore I strongly recommend *not*
installing version 2.02, but rather waiting until you receive release 3.0;
that is unless you want to install 2.02 because it will produce an
immediate benefit and raised productivity. Your current voice files for
NaturallySpeaking will transfer to version 2.02.<
This was in answer to a query about upgrading from Preferred to Deluxe vers
2.01 and it suggests that although voice files are transferable between
vers 2.01, this is not the case when upgrading from 2.01 to 3.0. Patrick
states that NatSpk 3.0 will need to be retrained from scratch. It follows,
therefore, that the critical voice files will not be preserved and since I
had intended to make the US English spelling my primary vocabulary, there
would have been no point in my installing v3.0 "on top" of 2.01 and even if
the files were not overwritten, to all intents and purposes they could be
uninstalled and the new version installed form scratch.
In other posts on the subject, I learned that it would be possible to
install both US and UK English versions but they would need to be installed
in different directories.
What I would like to know, from your (Holland's and others) experience is
this...
1. Could the ver 3.0 be installed to preserve the voice files etc of ver
2.02 and still allow them to be used (albeit without BestMatch adaptation)
but, in the meantime carry out training of new Topics according to
BestMatch?
2. Do you think ver 3.0 (US) version could be installed into a separate
folder from the existing ver 2.02 (English) without any conflicting *.dll
files, voice files vocabs?
The benefit here is that it would allow work to continue efficiently with
the old files but gradually implement the training with BestMatch until
such times as it's safe to switch completely to ver 3.0 (or not).
Look forward to any comments or enlightenment.
Best regards to all
Charles E. Horth
writing at 16:45 BST on 19 Jul 98
- Next message: Judy Evans: "Re: Is it just me?? and apology"
- Previous message: Lynda-L: "Correct That Command revisited"
- Next in thread: gNAt: "Re: Is it just me??"
 |  |