Vista Speech "Accuracy Slider"?

Does Vista Speech have a so-called "accuracy slider"
similar to the one in Dragon NaturallySpeaking?

Mark-

 

Yes.

--
..
--

"Mark Conrad" <none-of@your-business.invalid> wrote in message
news:250820091354498490%none-of@your-business.invalid...<!--coloro:blue--><span style="color:blue <!--/coloro-->
>
> Does Vista Speech have a so-called "accuracy slider"
> similar to the one in Dragon NaturallySpeaking?
>
> Mark- <!--colorc--><!--/colorc-->

 

In article <#TuY6TkJKHA.4652@TK2MSFTNGP04.phx.gbl>, <.> wrote:
<!--coloro:blue--><span style="color:blue <!--/coloro--><!--coloro:green--><span style="color:green <!--/coloro-->
> > Does Vista Speech have a so-called "accuracy slider"
> > similar to the one in Dragon NaturallySpeaking?<!--colorc--><!--/colorc-->
>
> Yes.<!--colorc--><!--/colorc-->


Where Andy, I can't find it?

Mark-

 

"Mark Conrad" <none-of@your-business.invalid> wrote in message
news:260820090519377757%none-of@your-business.invalid...<!--coloro:blue--><span style="color:blue <!--/coloro-->
> In article <#TuY6TkJKHA.4652@TK2MSFTNGP04.phx.gbl>, <.> wrote:
><!--coloro:green--><span style="color:green <!--/coloro--><!--coloro:darkred--><span style="color:darkred <!--/coloro-->
>> > Does Vista Speech have a so-called "accuracy slider"
>> > similar to the one in Dragon NaturallySpeaking?<!--colorc--><!--/colorc-->
>>
>> Yes.<!--colorc--><!--/colorc-->
>
>
> Where Andy, I can't find it?<!--colorc--><!--/colorc-->

I swear that has nothing to do with me my old friend, it is someone playing
silly buggers.

I have no reason to hide behind anything. You know this more than anyone
does. If I have got something to say, then I will say it.

andy t

 

"Mark Conrad" wrote:

<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Does Vista Speech have a so-called "accuracy slider"
> similar to the one in Dragon NaturallySpeaking?<!--colorc--><!--/colorc-->

Your question is not quite complete. In Dragon NaturallySpeaking, it is
called Speed vs. Accuracy Slider. WSR (Windows Speech Recognition) has
nowhere the overhead of Dragon NaturallySpeaking. On my machine, WSR takes up
90,000 kB. Dragon NaturallySpeaking takes up 296,000 kB, over three times
more resources. The purpose of the slider is for computers that have less RAM
and slower processors to make a decision as to whether they prefer to have
the words up here faster or more accurately.

For WSR, the words appear almost instantly on even Pentium 4 computers with
1 GB of RAM. There is no need for the Speed vs. Accuracy Slider.

Marty Markoe, eMicrophones, Inc.

 

That's not me either, I only refer you to Andy T.

--
..
--
<.> wrote in message news:%23TuY6TkJKHA.4652@TK2MSFTNGP04.phx.gbl...<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Yes.
>
> --
> .
> --
>
> "Mark Conrad" <none-of@your-business.invalid> wrote in message
> news:250820091354498490%none-of@your-business.invalid...<!--coloro:green--><span style="color:green <!--/coloro-->
>>
>> Does Vista Speech have a so-called "accuracy slider"
>> similar to the one in Dragon NaturallySpeaking?
>>
>> Mark- <!--colorc--><!--/colorc-->
><!--colorc--><!--/colorc-->

 

In article <965DE8F7-1733-47F5-A621-97CD8B7233B0@microsoft.com>,
<wwwMyMSSpeechcom@discussions.microsoft.com> wrote:
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Your question is not quite complete. <!--colorc--><!--/colorc-->

I know Martin, for the sake of brevity.

<!--coloro:blue--><span style="color:blue <!--/coloro-->
> The purpose of the slider is for computers that
> have less RAM and slower processors to make a
> decision as to whether they prefer to have
> the words up here faster or more accurately.<!--colorc--><!--/colorc-->

Correct.

The way it becomes "speedier" on a below-par
computer is to adjust the speech engine so it looks
at fewer words, next to the word it is trying to
comprehend.

Looking at fewer words-in-context risks WSR
guessing the wrong word, which is why the average
accuracy goes down.

Problem here is that not all speech engines are equal
in the sense that _some_ speech engines take longer
to figure out the correct word.


I can get 100% raw accuracy with WSR _provided_
that I speak in this fashion.

-------------------------
now is the time.....for all good men.....
to come to the aid of.... their country
-------------------------


At a speaking rate of say 96wpm.

Anything much faster than 96wpm causes the
raw accuracy to plummet.


Using the same high end computer, same approved
headset, same external sound card, same Vista OS,
with Dragon I can increase my speaking rate to 336wpm
and obtain the 100% raw accuracy.

Switching the OS to Apple's most recent version 10.5.8
and switching the speech app to "MacSpeech Dictate 1.5.2"
I can still speak at 336wpm with 100% raw accuracy.

(21 repetitions of the 16 word phrase:
"now is the time for all good men
to come to the aid of their country")

21 x 16 = 336wpm


I had some hope that WSR had a manual adjust slider
so I could play with the slider setting to optimize
the speed versus accuracy settings for WSR.

Hours of Googling failed to turn up any leads about
any such slider adjustment on WSR, so I assume
that there is no such slider.

Before you ask, all the Vista-side 'Speech' CP settings
funnel both input and output audio directly to my
headset, all other speakers, mics, etc are "disabled".


On the good side of this report, I had no problem adding
the complex words and getting consistant recognition:

nucleoside
gemcitabine
apoptosis
gastrointestinal
epithelium


I did have a lot of problems with stuff like:

PURPOSE/OBJECTIVE - - -(getting _quick_ uppercase)
radioresponse - - - (nospace command does not work here)
(25 mg/kg) - - - [ came out (25 MG/kilogram) ]

I expect I will find workarounds for these and other nuisances.


WSR seems to have more trouble with short words like "the"
and "aid" than the other two speech apps have.

"aid" spoken by itself is almost impossible for WSR to handle.

The other two speech apps have no trouble with "aid" or "the".

Mark-

--
I will keep plugging away for a full month, should be
interesting to see if I can get WSR performance closer to
what Dragon and MacSpeech are capable of.

 

> That's not me either, I only refer you to Andy T.

What do you mean by "I only refer you to Andy T?"

My name is andy t and andy is my real name. I have no reason, purpose or
desire to deviate from that name in these newsgroups and pretend to be
someone I am not, especially to someone like mark conrad whom I would want
to know it was I writing the post in answer to all his ridiculous, same OT
drivel.

Unlike the coward who shows himself as a Dot. It obviously has something to
hide. Either that or it cannot spell it's own name! (Must be a young
child.)

Also, if you look closely, I always sign my name or spell my name without
capitalisation.

As so; andy t, never Andy T.

andy t


<.> wrote in message news:eQmABAqJKHA.3724@TK2MSFTNGP03.phx.gbl...<!--coloro:blue--><span style="color:blue <!--/coloro-->
> That's not me either, I only refer you to Andy T.
>
> --
> .
> --
> <.> wrote in message news:%23TuY6TkJKHA.4652@TK2MSFTNGP04.phx.gbl...<!--coloro:green--><span style="color:green <!--/coloro-->
>> Yes.
>>
>> --
>> .
>> --
>>
>> "Mark Conrad" <none-of@your-business.invalid> wrote in message
>> news:250820091354498490%none-of@your-business.invalid...<!--coloro:darkred--><span style="color:darkred <!--/coloro-->
>>>
>>> Does Vista Speech have a so-called "accuracy slider"
>>> similar to the one in Dragon NaturallySpeaking?
>>>
>>> Mark-<!--colorc--><!--/colorc-->
>> <!--colorc--><!--/colorc--><!--colorc--><!--/colorc-->

 

"Mark Conrad" <none-of@your-business.invalid> wrote in message
news:260820092013288074%none-of@your-business.invalid...<!--coloro:blue--><span style="color:blue <!--/coloro-->
> In article <965DE8F7-1733-47F5-A621-97CD8B7233B0@microsoft.com>,
> <wwwMyMSSpeechcom@discussions.microsoft.com> wrote:
><!--coloro:green--><span style="color:green <!--/coloro-->
>> Your question is not quite complete.<!--colorc--><!--/colorc-->
>
> I know Martin, for the sake of brevity.<!--colorc--><!--/colorc-->

That's a first!
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Looking at fewer words-in-context risks WSR
> guessing the wrong word, which is why the average
> accuracy goes down.<!--colorc--><!--/colorc-->

I do not get that problem and as Marty's says, the words are on the page or
as soon as you say them, very accurate and much faster than Dragon. You do
not get the delay when dictating words in WSR as you do with Dragon.

Also, now using a 'Senneheiser MD431 ll Microphone' or on occasions the
'Senneheiser ME3 headset' and having the WSRToolkit on hand, I easily get up
to 99% accuracy now.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Problem here is that not all speech engines are equal
> in the sense that _some_ speech engines take longer
> to figure out the correct word.<!--colorc--><!--/colorc-->

Precisely right. Dragon takes longer and is slower than WSR to put dictated
words on the screen. I know this because I have tried both.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> I can get 100% raw accuracy with WSR _provided_ and
> that I speak in this fashion.<!--colorc--><!--/colorc-->

There is no such thing as 100% accuracy in ANY speech app, I don't care what
anybody says.

The human voice only has to dictate something out of character or at a
slightly different tone to normal for any speech app to interpret it
properly to the words you want. Thus, the occasional mistake will be there.

I will admit, I am not the fastest speaker in the world and I do like to
look back on what I have written, but it suits my purpose.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> -------------------------
> now is the time.....for all good men.....
> to come to the aid of.... their country
> -------------------------<!--colorc--><!--/colorc-->

Yeah, we know. Yawn!
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> (21 repetitions of the 16 word phrase:
> "now is the time for all good men
> to come to the aid of their country")<!--colorc--><!--/colorc-->

Yeah, we know. Yawn!
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> WSR seems to have more trouble with short words like "the"
> and "aid" than the other two speech apps have.
> "aid" spoken by itself is almost impossible for WSR to handle.<!--colorc--><!--/colorc-->

Funny, I do not get that problem.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> should be interesting to see if I can get WSR performance
> closer to what Dragon and MacSpeech are capable of.<!--colorc--><!--/colorc-->

I already can, and you would not call me a casual user, as I use it all the
time.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> I will keep plugging away for a full month,<!--colorc--><!--/colorc-->

Yeah, you keep plugging away. Perhaps one day you will learn how to use it
properly, hopefully, before you die.

andy t

 

In article <unnMPYyJKHA.4232@TK2MSFTNGP02.phx.gbl>, andy t
<none-of@your-businesseither.invalid> wrote:
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Yeah, you keep plugging away. Perhaps one day you will learn
> how to use it properly, hopefully, before you die.<!--colorc--><!--/colorc-->

Yes Andrew, although it is doubtful, being I already have one
foot in the grave, and the other foot on a slippery banana peel.

I am giving myself a full month to grok WSR, which
_should_ be long enough to overcome most of its foibles.

<!--coloro:blue--><span style="color:blue <!--/coloro-->
> I will admit, I am not the fastest speaker in the world...<!--colorc--><!--/colorc-->

Quite easy, really. I can speak four repititions of the
following 16 words within ten seconds:

now is the time for all good men
to come to the aid of their country


16 x 4 x 10 = 384wpm

Dragon recognizes the 384wpm speed with 100%
raw accuracy.

Naturally, I did not cheat by using any form of macro.

The only "trick" I used was a very slight pause after
the word "country", estimated at a tenth of a second.



MacSpeech Dictate 1.5.2 was not as speedy, I had to slow
down to about 256wpm to get my 100% accuracy.

So far, WSR is at the bottom of the speed ladder at 96wpm,
however I hope to jack that figure up as I gain more
experience with using WSR.

Now you should have no trouble at all squeezing in two
repetitions within ten seconds, a 192wpm speaking rate.

I doubt very much that you will be able to achieve 100%
raw accuracy at that speed, say every other time you try.

<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Also, now using a 'Senneheiser MD431 ll Microphone' or
> on occasions the 'Senneheiser ME3 headset' and having
> the WSRToolkit on hand...<!--colorc--><!--/colorc-->

I have a cheap Andrea NC-91 headset that I think came with
one of the Dragon apps.

I saved $15 by downloading the Microsoft macro-creation
file to roll my own macros.

<!--coloro:blue--><span style="color:blue <!--/coloro-->
> ...I easily get up to 99% accuracy now.<!--colorc--><!--/colorc-->

That is nice.


<!--coloro:blue--><span style="color:blue <!--/coloro-->
> There is no such thing as 100% accuracy in ANY speech app<!--colorc--><!--/colorc-->

I use accuracy figures to refer to _average_ accuracy, over
a period of time. If I speak at a modestly slow rate say 80wpm,
I have no difficulty filling up a page with no mistakes,
say two hundred words for example.

I guarantee if there are homonyms, then mistakes can occur,
and if a user tries to dictate similar short words in isolation,
such as an, and, end, then again mistakes are likely to occur,
no matter what microphone you happen to be using.

<!--coloro:blue--><span style="color:blue <!--/coloro-->
> ...and I do like to look back on what I have written...<!--colorc--><!--/colorc-->

Not me, it distracts my train of thought.


<!--coloro:blue--><span style="color:blue <!--/coloro-->
> ...and as Marty's says, the words are on the page or
> as soon as you say them...<!--colorc--><!--/colorc-->

Yep, distracting my thoughts.

Try to think constructively with a tribe of monkeys
jumping around inside your room.

<!--coloro:blue--><span style="color:blue <!--/coloro-->
> ...very accurate...<!--colorc--><!--/colorc-->

Not yet for me, I may change my opinion a month from now.

<!--coloro:blue--><span style="color:blue <!--/coloro-->
> ...and much faster than Dragon. <!--colorc--><!--/colorc-->

Not yet for me, 96wpm versus 384wpm.

I may change my opinion in a month, when I get more
experience running WSR.


<!--coloro:blue--><span style="color:blue <!--/coloro--><!--coloro:green--><span style="color:green <!--/coloro-->
> > Problem here is that not all speech engines are equal
> > in the sense that _some_ speech engines take longer
> > to figure out the correct word.<!--colorc--><!--/colorc-->
>
> Precisely right. Dragon takes longer and is slower
> than WSR to put dictated words on the screen.
> I know this because I have tried both.<!--colorc--><!--/colorc-->

Non-issue with me, as I seldom if ever look at the screen
when I am dictating, because it distracts my train of thought,
and interrupts the "flow" of those thoughts in my mind.


<!--coloro:blue--><span style="color:blue <!--/coloro-->
> The human voice only has to dictate something out of character
> or at a slightly different tone to normal for any speech app
> to interpret it properly to the words you want.
> Thus, the occasional mistake will be there.<!--colorc--><!--/colorc-->

I mostly agree, a stuffy nose from a cold will degrade accuracy
for example.


<!--coloro:blue--><span style="color:blue <!--/coloro--><!--coloro:green--><span style="color:green <!--/coloro-->
> > -------------------------
> > now is the time.....for all good men.....
> > to come to the aid of.... their country
> > -------------------------<!--colorc--><!--/colorc-->
>
> Yeah, we know. Yawn!<!--colorc--><!--/colorc-->

One interesting thing (to me) is where I broke the word "aid".

Earlier I tried to break it thus:

to come to the aid... of their country

(notice the subtle difference in where I paused)

ALL SORTS of mistakes occurred, such as:

"to come to the eight of their country"

- - - or - - -

"to come to the aide of their country"

No amount of correction would deter WSR from making
those particular mistakes, I tried everything except
macros, which I hate to use.


Seems that "aid" needed the words preceeding it,
in order to figure out the correct word.

(dull subject, I agree)

Mark-

 

FWIW, the reason I am devoting a full month to learning
the ins and outs of WSR, is because I really hope to be
able to use it, if for no other reason that to apply a little
pressure to Nuance, to get them to act more compassionate
towards their own customer base.

The present owner of Dragon is Nuance, a _very_
bloodthirsty international speech business,
based in Belgium.

Nuance could teach Microsoft dirty tricks.

Like it or not, Nuance "owns" speech recognition in the USA.

I have yet to find any large hospital here that uses
anything except Dragon NaturallySpeaking.

The version 10.0 full medical at $1,600 has a formidable
300,000 word built-in vocabulary, which can be added to
by the user with even more words, if desired.

Now WSR is more than adequate for "everyday speech"
and hobby use, but it will *NEVER* be able to compete
with Dragon for business uses, such as medical or legal.

WSR was never designed to do so, as Marty has stated.



All that said, I certainly would like to get WSR to perform
better for me than it presently does, mainly as regards
dictation speed and _quick_ and efficient text correction
features like Dragon has.

As regards my wish list, I wish the new "Airline 77"
wireless microphone from Samson was two-way.

In other words could hear original audio playback from
Dragon. Unfortunately it is a microphone only, with no
provision for wireless playback of original audio
from the Dragon app.

New feature of the newest Airline 77 is that it operates
in the 642MHz to 646MHz range.
(not tunable in that range, user has to select one of
six available frequencies BEFORE he buys)

This is of importance in my case, because the local hospital
is close enough to a digital TV station here to impose serious
interference to the older Airline 77 model, which operates
in the same 800 MHz range as the new digital TV station.

Enough rambling, back to trying to grok WSR.

Mark-

 

"Mark Conrad" <none-of@your-business.invalid> wrote in message
news:270820091802596093%none-of@your-business.invalid...<!--coloro:blue--><span style="color:blue <!--/coloro-->
> In article <unnMPYyJKHA.4232@TK2MSFTNGP02.phx.gbl>, andy t
> <none-of@your-businesseither.invalid> wrote:<!--colorc--><!--/colorc-->
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Quite easy, really. I can speak four repititions of the
> following 16 words within ten seconds:<!--colorc--><!--/colorc-->

Repetitions, not *repititions*.

I suppose now you are going to Tell me that you did not use Dragon to
dictate that word. Hmm... Figures!

You see, you are being smug again and this is what happens, you make
yourself look a fool once more. Straight away, this dismisses your 100%
accuracy theory with Dragon.

This is exactly what I am trying to say. You only need to pronounce a word
slightly incorrectly, especially at speed, for either the wrong word to come
up or it gets confused and spells the word incorrectly, as you have done
above.

Once more for the record. Speech apps can and have been known to spell
words wrong on the screen. It has happened to me with WSR and Dragon.

You say you do not use Dragon all the time. Why do you not try and use it
all the time and use words that are not written down for you. By this I
mean words that come into your head as you would dictate normally as I have
to, see how many mistakes you make then.

Why would I want to dictate a set of words already set down on a piece of
paper? If I wanted to do that, I would simply cut and paste them into a
document. That is far quicker than using Dragon or WSR for that matter.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> now is the time for all good men
> to come to the aid of their country
>
>
> 16 x 4 x 10 = 384wpm<!--colorc--><!--/colorc-->

Um, repetition, not dictation.

If you recite certain words in order over and over again, any speech app
will remember those words and so it is not such a great feat. I can do this
no problem with WSR but that is not why I use it.

I use it as it was intended for, to be used for proper dictating and proper
dictating means the words are ad-libbed and come from your head, not a piece
of paper. (Repetitiveness).

Nobody normally, narrates like that. Imagine if you were writing something,
you would always stop and look back at each line or paragraph from time to
time. Secondly, you are narrating from a set script, which is unnatural
unless you write out what you want to dictate first, but that would be
defeating the object and it would be slower, not quicker.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Dragon recognizes the 384wpm speed with 100%
> raw accuracy.<!--colorc--><!--/colorc-->

That proves nothing and does not necessarily mean it is 100% accurate. I
reiterate, none of the speech apps at this moment in time are 100% accurate
when using dictation properly and that does not mean reading from a script.
I have explained this already. (Repetitiveness).

I have tried WSR without any scripts but just speaking fast with whatever
comes into my head. There were no mistakes apart from the odd word, it
could not decipher because of the speed I was dictating, thus pronouncing
the odd word incorrectly. This happens because as I say, that is not the
natural way to dictate. Any speech app including Dragon, will only respond
to whatever the human voice feeds it and as we are only human, dictating at
that speed will inevitably bring up mistakes.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Naturally, I did not cheat by using any form of macro.<!--colorc--><!--/colorc-->

I only use text macros in special circumstances such as to put my name at
the bottom of a letter or document by saying the words 'sign off' (I believe
Marty uses this macro as well for his own name and address), or by putting
in my address by saying the words 'address'. I will agree with you on that
one, the less text macros the better.

Command macros are of different kettle of fish. For someone like me who is
disabled, they are a godsend. I can zip through emails, expand or collapse
folders in an instant, such as in the newsgroups and lots more. As I say,
for someone such as me, command macro's are a necessity. Also, it is no
problem at all if I forget one of them. It is easy enough to look them up
in the speech macros folder.

Plus WSR has the 'show numbers' command, which can pick out any webpage,
e-mail or any point on the screen in an instant and is much faster than any
command Dragon possesses except possibly a purpose made command macro.
Perhaps having this facility has made me lazy, but with it, I can get to
anywhere around a computer with this much faster than in Dragon.

Having said all that, in the very near future, a local Charity have kindly
put up the funding so I may have Dragon Professional and a certain amount of
lessons on it. This may give me the incentive I need to have the choice.

I have also ordered Windows 7 when it comes out in October. You were right
about that, there are some vast improvements within the interface regarding
Windows speech recognition.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> So far, WSR is at the bottom of the speed ladder at 96wpm,
> however I hope to jack that figure up as I gain more
> experience with using WSR.<!--colorc--><!--/colorc-->

As I said before, natural dictation is not all about speed. If you can
think what you're going to say as in ad-libbing and putting that to the
screen at that speed without any mistakes, then you're a better man than I
Gunga Din. Again, the way you are showing your examples, it is not natural
dictation and no test at all. It proves nothing. (Repetitiveness).
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Now you should have no trouble at all squeezing in two
> repetitions within ten seconds, a 192wpm speaking rate.<!--colorc--><!--/colorc-->

Um, once again, that is not dictation. You even mention it yourself above
as a repetition. The speech app learns the words in the order you are
saying them. That is not true dictation. (Repetitiveness).
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> I have a cheap Andrea NC-91 headset that I think came with
> one of the Dragon apps.<!--colorc--><!--/colorc-->

Then I suggest you use or buy a decent microphone and then use it with WSR.

If you remember rightly, I started by using only the integral Mic within my
laptop for WSR and I will admit, I was probably getting only about 80%
accuracy. Since having this new Senneheiser, my dictation has shot up to as
near to 100% as you can get. I still get the odd mistake, but I do with
Dragon 10 also.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> I saved $15 by downloading the Microsoft macro-creation
> file to roll my own macros.<!--colorc--><!--/colorc-->

If it is the original WSR speech macro book written by Brad T, it is a very
handy thing to have if you are using WSR. You should try putting some of
your dictated documents through the 'Add from file' section. By doing this,
the words are analysed by the way you speak them. Or, put some of them in
the 'Train from text' section and train your speech recognition some more.
This will also add to the accuracy level. I got it for nothing because I am
disabled ;-)

Once I did this, I could not believe the accuracy level. It suited me fine
and the more you put in, the more ammunition it has to work with. Thus, the
accuracy level will rise also. You have also been advised before to add
words like 'aid' or bunch of words involving that word into the speech
dictionary. This is another way of training and for WSR to recognise
certain words. As I said before, I have no problem dictating the word
'aid'.

I believe I need first aid. I believe I need some aid in learning. I hope
he will come to my aid.

I just dictated those three sentences in about 10 seconds straight off with
no cheating, all with the word 'aid' in.
<!--coloro:blue--><span style="color:blue <!--/coloro--><!--coloro:green--><span style="color:green <!--/coloro-->
>> There is no such thing as 100% accuracy in ANY speech app<!--colorc--><!--/colorc-->
>
> I use accuracy figures to refer to _average_ accuracy, over
> a period of time. If I speak at a modestly slow rate say 80wpm,
> I have no difficulty filling up a page with no mistakes,
> say two hundred words for example.<!--colorc--><!--/colorc-->

I think we are agreeing on the same thing here. There is no such thing as
100% speech accuracy in ANY speech app, no matter how you dress it up.
(Repetitiveness).
<!--coloro:blue--><span style="color:blue <!--/coloro--><!--coloro:green--><span style="color:green <!--/coloro-->
>> ...and I do like to look back on what I have written...<!--colorc--><!--/colorc-->
>
> Not me, it distracts my train of thought.<!--colorc--><!--/colorc-->

By reading the same copied out text repeatedly at a fast pace as if it's a
competition, does not constitute towards your thoughts being distracted; a
child of 10 could do that. Again, no one dictates like that.
(Repetitiveness).
<!--coloro:blue--><span style="color:blue <!--/coloro--><!--coloro:green--><span style="color:green <!--/coloro-->
>> ...and as Marty's says, the words are on the page or
>> as soon as you say them...<!--colorc--><!--/colorc-->
>
> Yep, distracting my thoughts.<!--colorc--><!--/colorc-->

What thoughts? If the words are already on a piece of paper and you are
reciting them over and over again repetitively, you are not using any
thoughts that come from the mind because they are there in front of you.
That is hardly thinking it is.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Try to think constructively with a tribe of monkeys
> jumping around inside your room.<!--colorc--><!--/colorc-->

I fail to see how it can be constructive if all you're doing is reading out
copied text repeatedly at speed, like a rabid monkey.

Thinking constructively takes time and care. In the end, the finished
product that I see on the screen, is precisely what I want to see, not some
useless copied out text that I could cut and paste much quicker anyway.
(Repetitiveness).

It is what my brain wants to see, not some useless text my mouth just spouts
out copied from somewhere else.
<!--coloro:blue--><span style="color:blue <!--/coloro--><!--coloro:green--><span style="color:green <!--/coloro-->
>> ...very accurate...<!--colorc--><!--/colorc--><!--colorc--><!--/colorc-->
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Not yet for me, I may change my opinion a month from now.<!--colorc--><!--/colorc-->

Good luck but WSR is accurate enough for me and definitely faster than
Dragon to the screen. I hate the delay that Dragon possesses when
dictating. Perhaps that is something that Nuance can address in the future,
I may be inclined to try some more with it if they did.
<!--coloro:blue--><span style="color:blue <!--/coloro--><!--coloro:green--><span style="color:green <!--/coloro-->
>> ...and much faster than Dragon.<!--colorc--><!--/colorc-->
>
> Not yet for me, 96wpm versus 384wpm.<!--colorc--><!--/colorc-->

I have already explained that. It is not a competition or speed test. You
are beginning to repeat yourself again Mark and I am beginning to repeat
myself repeating the same answers. At this rate, I will become as insane as
you. (Repetitiveness).

Once again I am getting that feeling of déjà vu. You remember that word
don't you Mark, the French word that you didn't know the meaning of.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
>I may change my opinion in a month, when I get more
>experience running WSR.<!--colorc--><!--/colorc-->

As I've said in the past, your opinion to me is about as important as a flea
farting and means nothing.

Why are you bothering anyway when you insist that your Dragon is far
superior? Stick with Dragon if it is your cup of tea.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Non-issue with me, as I seldom if ever look at the screen
> when I am dictating,<!--colorc--><!--/colorc-->

You don't need to look at the screen if all you are doing is reading out
copied text repetitively, from a piece of paper. Once again, that is not
the way it was meant to be used. (Repetitive).
<!--coloro:blue--><span style="color:blue <!--/coloro-->
>because it distracts my train of thought,
> and interrupts the "flow" of those thoughts in my mind.<!--colorc--><!--/colorc-->

You mean the thoughts that are written down on paper first and then
repetitively spoken in an endless loop as though you were reading from the
Koran. (Repetitive).
<!--coloro:blue--><span style="color:blue <!--/coloro--><!--coloro:green--><span style="color:green <!--/coloro-->
>> The human voice only has to dictate something out of character
>> or at a slightly different tone to normal for any speech app
>> to interpret it properly to the words you want.
>> Thus, the occasional mistake will be there.<!--colorc--><!--/colorc-->
>
> I mostly agree, a stuffy nose from a cold will degrade accuracy
> for example.<!--colorc--><!--/colorc-->

Or early in the morning, for example when the vocal cords can sound
different from later on in the day. They human voice is not a machine.
After all, it is only human. Unlike a speech app, which is machine
orientated.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
>
><!--coloro:green--><span style="color:green <!--/coloro--><!--coloro:darkred--><span style="color:darkred <!--/coloro-->
>> > -------------------------
>> > now is the time.....for all good men.....
>> > to come to the aid of.... their country
>> > -------------------------<!--colorc--><!--/colorc--><!--colorc--><!--/colorc-->
>
> One interesting thing (to me) is where I broke the word "aid".
>
> Earlier I tried to break it thus:
>
> to come to the aid... of their country
>
> (notice the subtle difference in where I paused)
>
> ALL SORTS of mistakes occurred, such as:
>
> "to come to the eight of their country"
>
> - - - or - - -
>
> "to come to the aide of their country"<!--colorc--><!--/colorc-->

Um, written down repetitiveness. Not natural dictation.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Seems that "aid" needed the words preceeding it,
> in order to figure out the correct word.<!--colorc--><!--/colorc-->

I have already explained. (Repetitiveness).

I do not have a problem with that word.

It does seem however that your Dragon has a problem spelling the word
'preceding' as you have shown above, once again proving that your speech app
is not 100% accurate, no speech app is.


andy t

 

"Mark Conrad" <none-of@your-business.invalid> wrote in message
news:270820092353301398%none-of@your-business.invalid...<!--coloro:blue--><span style="color:blue <!--/coloro-->
>
>
> The version 10.0 full medical at $1,600 has a formidable
> 300,000 word built-in vocabulary, which can be added to
> by the user with even more words, if desired.<!--colorc--><!--/colorc-->

So what! I can do exactly the same with WSR as I can do with Dragon. Plus
I get a whole operating system with it at half the price that Dragon medical
costs.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Now WSR is more than adequate for "everyday speech"
> and hobby use, but it will *NEVER* be able to compete
> with Dragon for business uses, such as medical or legal.<!--colorc--><!--/colorc-->

Never say never!

BTW, you are retired and you use Dragon for your use and your hobby. I use
WSR out of necessity, so it is no hobby of mine.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> All that said, I certainly would like to get WSR to perform
> better for me than it presently does, mainly as regards
> dictation speed and _quick_ and efficient text correction
> features like Dragon has.<!--colorc--><!--/colorc-->

I told you before, I have WSR and Dragon 10 and I find WSR easier to use and
quicker for my needs.
Maybe when I get my Dragon professional and Windows 7, I may use both then
decide which one is best for me.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> As regards my wish list, I wish the new "Airline 77"
> wireless microphone from Samson was two-way.<!--colorc--><!--/colorc-->

The Airline 77 is the poor man's Senneheiser 431 ll and in all the read ups
I have seen, the critics do not rate it all that much and that it suffers
from air frequency problems which means it picks up a lot of background
noise. This makes it not a great microphone for accuracy. That may well be
why it is cheap.

andy t

 

In article <ORmqJJNKKHA.4316@TK2MSFTNGP04.phx.gbl>, andy t
<none-of@your-businesseither.invalid> wrote:
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Maybe when I get my Dragon professional and Windows 7,
> I may use both then decide which one is best for me.<!--colorc--><!--/colorc-->

That's the spirit, test them both.

Keep in mind that it will be just as tough for you
to give a fair test to Dragon, as it is for me to give a
fair test to WSR.

You are used to WSR, and I am used to Dragon.

Complicating my situation is that I am spending
a lot of time with "MacSpeech Dictate", so I am
starting to neglect both Dragon and WSR.


MacSpeech Dictate is so buggy that you can not even
imagine how a guy like me tolerates it.

Reason of course is speed, it can handle my dictation
as fast as I normally speak, say about 160wpm.

WSR at the present time is performing at half that,
even with your so-called "natural dictation".


<!--coloro:blue--><span style="color:blue <!--/coloro--><!--coloro:green--><span style="color:green <!--/coloro-->
> > I have a cheap Andrea NC-91 headset that I think
> > came with one of the Dragon apps.<!--colorc--><!--/colorc-->
>
> Then I suggest you use or buy a decent microphone and
> then use it with WSR.<!--colorc--><!--/colorc-->

Waste of money, same headset performs accurately with
Dragon at 336wpm, therefore it can't be the headset.

Biggest waste of money on my part was when I fell for
the "magic microphone" nonsense, wasting hundreds of
dollars on very expensive Sennheiser models.

Eventually I will find the quirk in WSR that is
responsible for _my_ WSR not recognizing the
word "aid" when it is spoken in isolation.

....and it will _not_ be the cheap headset that is
responsible for the problem.

Mark-

 

<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Once more for the record. Speech apps can and have been
> known to spell words wrong on the screen. It has happened
> to me with WSR and Dragon.<!--colorc--><!--/colorc-->

Has not happened to me in 25 years of using speech apps
of all brands, including WSR.

According to "Chucker" in the KnowBrainer forum, it is
impossible for that to occur, as the words in the built-in
vocabulary are taken from regular recognized dictionaries.

Chucker is a recognized authority on SR issues, having
worked for Dragon as a programmer for many years,
running an entire department of other programmers.

I have never known of a speech app right-out-of-the-box
misspelling a word, including WSR, which is overtly prone
to outputting words by merely breathing on the microphone,
unlike more mature speech apps like Dragon.

Speech apps can *ONLY* output the words that are in
their dictionary, or in a specially crafted macro, according
to the speech recognition expert "Chucker", a moderator in
the KnowBrainer forum.

I can see all 300,000 built-in words in Dragon, which is
impossible with WSR, as it only shows user-created words,
refusing to output a complete listing of all its default
built-in words.



Common slang words are included in all speech apps.

A user can _add_ a deliberately misspelled word.

For example with Dragon, I can add the deliberately
misspelled word "repititions", then delete the correct built-in
word "repetitions".

In that case I can force Dragon to misspell a word, however I
had to take overt action to force the misspelling.

Mark-

 

In article <#UvODxMKKHA.4232@TK2MSFTNGP02.phx.gbl>, andy t
<none-of@your-businesseither.invalid> wrote:
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Why would I want to dictate a set of words...<!--colorc--><!--/colorc-->

As a speed test, unless of course you are afraid to compare
the speed of WSR against other speech apps.


<!--coloro:blue--><span style="color:blue <!--/coloro-->
> ...already set down on a piece of paper?<!--colorc--><!--/colorc-->

now is the time for all good men
to come to the aid of their country

Are you incapable of memorizing that short sentence?


<!--coloro:blue--><span style="color:blue <!--/coloro-->
> If I wanted to do that, I would simply cut and paste
> them into a document.<!--colorc--><!--/colorc-->

That would not get your comparative speed test done,
which you are apparently afraid to do.


<!--coloro:blue--><span style="color:blue <!--/coloro--><!--coloro:green--><span style="color:green <!--/coloro-->
> > now is the time for all good men
> > to come to the aid of their country
> >
> >
> > 16 x 4 x 10 = 384wpm<!--colorc--><!--/colorc-->
>
> Um, repetition...<!--colorc--><!--/colorc-->

Right, four repetitions to be precise.


<!--coloro:blue--><span style="color:blue <!--/coloro--><!--coloro:green--><span style="color:green <!--/coloro-->
> > now is the time for all good men
> > to come to the aid of their country
> >
> >
> > 16 x 4 x 10 = 384wpm<!--colorc--><!--/colorc-->
>
> ...not dictation.<!--colorc--><!--/colorc-->

Wrong.

dictation (noun)
The action of saying words aloud to be typed,
written down, or recorded on tape.


Last time I checked, my dictation was "written down"
into a computer file.




<!--coloro:blue--><span style="color:blue <!--/coloro-->
> I use it as it was intended for, to be used for proper
> dictating and proper dictating means the words
> are ad-libbed and come from your head, not a
> piece of paper.<!--colorc--><!--/colorc-->

Who said anything about a piece of paper, you originated
that idea as a straw man issue to deflect attention away
from a speed test which you are afraid to perform.


<!--coloro:blue--><span style="color:blue <!--/coloro-->
> I use it as it was intended for, to be used for proper
> dictating and proper dictating means the words
> are ad-libbed and come from your head...<!--colorc--><!--/colorc-->

Bull****, that is just one narrow use of speech
recognition, I do not need some self-appointed
dictator to tell me what I can or can't do with my
speech recognition app.


<!--coloro:blue--><span style="color:blue <!--/coloro-->
> If you recite certain words in order over and over
> again, any speech app will remember those words...<!--colorc--><!--/colorc-->

Well, yes and no.

Speech app will merely consider the _probability_
of words occurring next to other very nearby words.

...come to the aid of...

- - - versus - - -

...come to the aide of...


Where both words sound similar, and the speech app
has to make a choice between the two versions.





If I instead had dictated:

...come to the start of....


Then *NO* *WAY* would the speech app output

...come to the aid of...



....no matter how many hundreds of times I had
dictated ...come to the aid of... previously.


Mark-

 

In article <#UvODxMKKHA.4232@TK2MSFTNGP02.phx.gbl>, andy t
<none-of@your-businesseither.invalid> wrote:
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> ...natural dictation is not all about speed.<!--colorc--><!--/colorc-->

I disagree, as do the vast majority of others.

Speech recognition is a marginal technology.

Take away the speed aspect, then there is not
much left for the usual computer user.

He/she might as well revert to using the
computer keyboard.


<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Again, the way you are showing your examples,
> it is not natural dictation...<!--colorc--><!--/colorc-->

It is not intended to be natural dictation, it is
intended to be a comparative speed test.

If WSR was only capable of 10wpm without gross
mistakes, would you still use it?

(YOU probably would) ;-)


<!--coloro:blue--><span style="color:blue <!--/coloro-->
> and no test at all. It proves nothing.<!--colorc--><!--/colorc-->

96wpm versus at least 336wpm, it "proves"
that WSR is at least 3 times slower than Dragon.

That 96wpm is really pushing it, I feel a lot better
dictating at a slower 80wpm with WSR.

Time for correcting text would knock that down to
about 40wpm.

Speed is all important in business, given the
extremely high cost of computers, speech apps,
and the IT staff to keep everything running.

Without speed, they would be better off hiring
fast typists and running cheap computers.

Hobby speech uses are a very minor part of
speech recognition sales.

One of the reasons that medical care in the USA
is on the verge of breaking the economy,
trillions of dollars, is because of the high cost of
handling the paperwork associated with treatment.

Return-on-investment study after study after study
has shown that speech recognition, properly
implemented, can save great gooey gobs of money,
compared to a typing staff of trained monkeys
banging away on keyboards.

The computers _have_ to be high end costly beasts.

(my MacBook Pro gets so hot that I typically fry my
breakfast eggs on it when I am running my $1,600
Dragon medical app in its Vista partition on my Mac)

The speech apps themselves have to be costly, and that is
not merely for the vocabulary. It is well known that
Dragon has much more expansive and _quicker_ text
correction features than a hobby speech app like WSR.

Speed, speed, speed, that is the story.

....and expense. Nuance and indirectly Microsoft
benefit big time from this money being spent
on medical crap.

For practical reasons, if you run a hospital, you have
to hire an expensive IT staff to keep the costly PCs
running, and to train the hospital staff about how to use
the expensive software, such as the EMR software that
works with Dragon. (EMR = Electronic Medical Record)

EMR software can easily run $5,000

....such as the "MacPractice MD" software for Macs,
that is used to fill out medical EMRs by voice, and
includes handling MRI scans, CAT scans, and provision
for "talking to" the PC-only EMR forms used by
Windows-using doctors and hospitals.

Windows presently has the speech market sewed up.

Macs probably get less that one penny of every dollar
that outfits like Nuance and Microsoft get.

Mark-

 

In article <290820092021580244%none-of@your-business.invalid>, Mark
Conrad <none-of@your-business.invalid> wrote:
<!--coloro:blue--><span style="color:blue <!--/coloro--><!--coloro:green--><span style="color:green <!--/coloro--><!--coloro:darkred--><span style="color:darkred <!--/coloro-->
> > > I have a cheap Andrea NC-91 headset that I think
> > > came with one of the Dragon apps.<!--colorc--><!--/colorc-->
> >
> > Then I suggest you use or buy a decent microphone and
> > then use it with WSR.<!--colorc--><!--/colorc-->
>
> Waste of money, same headset performs accurately with
> Dragon at 336wpm, therefore it can't be the headset.
>
> Biggest waste of money on my part was when I fell for
> the "magic microphone" nonsense, wasting hundreds of
> dollars on very expensive Sennheiser models.
>
> Eventually I will find the quirk in WSR that is
> responsible for _my_ WSR not recognizing the
> word "aid" when it is spoken in isolation.
>
> ...and it will _not_ be the cheap headset that is
> responsible for the problem.<!--colorc--><!--/colorc-->

I finally solved the problem of the word "aid" not being
correctly recognized, whenever it was spoken by itself.

Was a corrupted WSR profile, _not_ the cheap headset.

WSR fails to offer control over when or even if
a users profile gets saved.

If a users cat vocalizes a husky meow, or someone farts
or belches, those extraneous noises have the potential
to corrupt a users profile

By contrast, with Dragon _or_ MacSpeech, I have full
control over *if* my profile gets saved, so can I can
choose to _not_ save my speech profile after a
dictation session, should I desire.

Mark-

 

In article <300820090814330751%none-of@your-business.invalid>, Mark
Conrad <none-of@your-business.invalid> wrote:
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> I finally solved the problem of the word "aid" not being
> correctly recognized, whenever it was spoken by itself.
>
> Was a corrupted WSR profile, _not_ the cheap headset.
>
> WSR fails to offer control over when or even if
> a users profile gets saved.
>
> If a users cat vocalizes a husky meow, or someone farts
> or belches, those extraneous noises have the potential
> to corrupt a users profile.<!--colorc--><!--/colorc-->

Okay, now I am making real progress with the flaky WSR.

Ran the speed test over again with the fresh profile.

Kicked out the cat so it would not meow, taped up my
anal orifice, guzzled some anti-belch tablets, unplugged
the refrigerator so it would not come on, had a friend
stop traffic on the road outside my house, and took
various other precautions to prevent the flaky WSR from
hearing any sounds except the pear-shaped tones
coming from my mouth.

Sheesh! - none of this nonsense is necessary with Dragon,
as I can even have the TV on when I am dictating.
(same cheap microphone, BTW)

Okay, speed performance jumped up to 192wpm.

That was 12 repetitions in 60 seconds, 100% accuracy
on all 192 words.
(MacSpeech is 256wpm, Dragon 336wpm)

192wpm is quite amazing for a free speech app,
in my opinion.

Was unable to achieve 100% raw accuracy when I increased
the dictation speed, however invariably regained 100%
accuracy whenever I decreased dictation
speed down to 192wpm, on the test sentence:

now is the time for all good men
to come to the aid of their country

Mark-

 

"Mark Conrad" <none-of@your-business.invalid> wrote in message
news:290820092022271984%none-of@your-business.invalid...<!--coloro:blue--><span style="color:blue <!--/coloro-->
> In article <#UvODxMKKHA.4232@TK2MSFTNGP02.phx.gbl>, andy t
> <none-of@your-businesseither.invalid> wrote:
><!--coloro:green--><span style="color:green <!--/coloro-->
>> ...natural dictation is not all about speed.<!--colorc--><!--/colorc-->
>
> I disagree, as do the vast majority of others.
>
> Speech recognition is a marginal technology.
>
> Take away the speed aspect, then there is not
> much left for the usual computer user.
>
> He/she might as well revert to using the
> computer keyboard.<!--colorc--><!--/colorc-->

Bull****! I can dictate much faster than using a keyboard and think at the
same time as to what I want to write instead of writing or taping the words
first. What a waste of time that is.

<!--coloro:blue--><span style="color:blue <!--/coloro-->
> 96wpm versus at least 336wpm, it "proves"
> that WSR is at least 3 times slower than Dragon.<!--colorc--><!--/colorc-->

The same old words are over and over again. Boy, that will come in handy.
<!--coloro:blue--><span style="color:blue <!--/coloro-->
> That 96wpm is really pushing it, I feel a lot better
> dictating at a slower 80wpm with WSR.<!--colorc--><!--/colorc-->

You would say that, being an agent of Dragon. I find WSR much easier to use
and the end result is the same.

<!--coloro:blue--><span style="color:blue <!--/coloro-->
> Speed, speed, speed, that is the story.<!--colorc--><!--/colorc-->

You mean that is your story. You must be taking speed, speed, speed. On
and on and on and on.

andy t