Re: Question for Windows Speech Recognition guru

Mark Conrad wrote:
> Need advice from Vista Speech Recognition guru.
> (formal name WSR, for Windows Speech Recognition)
>
> The "Analyze Written Documents" feature of WSR
> appears to be broken.
>
> I try to keep up-to-date on three different SR apps,
> which is not easy, because of the differences.
>
> The three SR apps are:
>
> 1) Dragon Pro
> 2) WSR (Vista Speech Recognition)
> 3) MacSpeech Dictate
>
> ...using most recent versions of above 3 apps
>
>
> Without going into the reason _why_ , I would
> like to be able to speak the following, and
> have it come out exactly as shown in
> the two examples below:
>
> - - - examples - - - examples - - -
> ---------------------------------------
> the three words cent scent sent are homonyms
>
> the four words two too two tutu are homonyms
> ---------------------------------------
>
>
> This works fine with both Dragon and MacSpeech,
> but not with WSR (Windows Speech Recognition)
>
>
> Anyone manage to get the "analyze written documents"
> feature of WSR to work on the above two examples?
>
> Mark-

First of all, as an English major from long ago, I must tell you that
tutu is not a homonym of to, too, and two.
Your second example repeats the word two twice. I'm amazed that Dragon's
analysis of the two sentences you cite actually works. I'd expect WSR
could pick up the difference between to, too, and two when each is used
in context, such as "He went to the store to buy two bottles of beer.
She did, too." or "Willy sent Millicent a silly scent."

Your examples do not use the words in context.

Bill

"Bill Sharpe" <> wrote in message
news:...
> Mark Conrad wrote:
>> Need advice from Vista Speech Recognition guru.
>> (formal name WSR, for Windows Speech Recognition)
>>
>> The "Analyze Written Documents" feature of WSR appears to be broken. I
>> try to keep up-to-date on three different SR apps,
>> which is not easy, because of the differences.
>>
>> The three SR apps are:
>>
>> 1) Dragon Pro
>> 2) WSR (Vista Speech Recognition)
>> 3) MacSpeech Dictate
>>
>> ...using most recent versions of above 3 apps
>>
>>
>> Without going into the reason _why_ , I would
>> like to be able to speak the following, and have it come out exactly as
>> shown in the two examples below:
>>
>> - - - examples - - - examples - - -
>> ---------------------------------------
>> the three words cent scent sent are homonyms
>>
>> the four words two too two tutu are homonyms
>> ---------------------------------------
>>
>>
>> This works fine with both Dragon and MacSpeech, but not with WSR (Windows
>> Speech Recognition)
>>
>>
>> Anyone manage to get the "analyze written documents" feature of WSR to
>> work on the above two examples?
>>
>> Mark-
>
> First of all, as an English major from long ago, I must tell you that tutu
> is not a homonym of to, too, and two.
> Your second example repeats the word two twice. I'm amazed that Dragon's
> analysis of the two sentences you cite actually works. I'd expect WSR
> could pick up the difference between to, too, and two when each is used in
> context, such as "He went to the store to buy two bottles of beer. She
> did, too." or "Willy sent Millicent a silly scent."
>
> Your examples do not use the words in context.
>
> Bill

There you go Mark, now you have learned something and that 'tutu' is not a
homonym. Thought you would know that.

I have just dictated the two lines below using WSR all in context. WSR, as
you can see can decipher this and knows which one to use and where to put
them. The first one used the number 2 as opposed to 'two' but the second
one actually used the written 'two' but in each case, because they are
written in context, WSR can fathom out where to use them as in 'to', 'two'
(or sometimes 2) and 'too'.


The second 2 identical lines below the first two, I have dictated with
Dragon. You will notice that Dragon has more of a problem than WSR does in
picking out the written form 'two' as opposed to the numeral 2.

I am going to have 2 cups of tea, let's hope they are not too hot.

I am going to the pub, I will have at least two pints and so lets hope they
are not too warm.

I am going to have 2 cups of tea, let's hope they are not too hot.

I am going to the pub, I will have at least 2 pints and so lets hope they
are not too warm.

andy t

Mark Conrad wrote:
>> There you go Mark, now you have learned something
>> and that 'tutu' is not a homonym.
>
> It certainly is. Below is a list of two homonyms:
>
> tutu tutu
>
>
>
> Here is another list of two homonyms:
>
> tutu two
>
>
> homonym (noun)
> each of two words having the same pronounciation
> but different meanings, origins, or spelling
>
> list (noun)
> a number of connected items written or printed
> consecutively
>
> Mark-

Do you really believe that "tutu," a two-syllable word has the same
pronunciation as "two," a one-syllable word?

Bill

Mark Conrad wrote:
> the four words two too two tutu are homonyms
other material snipped
> Mark-

Just because you speak such a phrase doesn't make it so. Plus there are
only three different versions of to in your sentence.

Try speaking the following four sentences one after another and see if
your speech recognition software gets them all in the same order:

The three words to too and two are homonyms.
The three words too to and two are homonyms.
The three words two to and two are homonyms.
The three words two too and to are homonyms.

If you want to take the "and" out, that's fine, but take it out of all
four sentences. The context may be sufficient to get all three versions,
but I doubt very much that the order will remain as above. And, to be
fair, the order really doesn't matter.

Bill

See my inline comments.

Bill

Mark Conrad wrote:
> In article <>, Bill Sharpe
> <> wrote:
>
>> Do you really believe that "tutu," a two-syllable word
>> has the same pronunciation as "two," a one-syllable word?
>
> Of course not, I never said they did have the same
> pronunciation.

Well, then, tutu is not a homonym of two.
>
> Consider this, if I say:
>
> "the four words two two two two two are homonyms"
>
> WHERE is my word "tutu" located?
>
> You have four choices:
>
> tutu two two two ?
> two tutu two two ?
> two two tutu two ?
> two two two tutu ?
>
> You would not be able to tell which of the four
> choices until the result _I_ selected ahead
> of time was printed, would you?

True enough, but what do you mean by the result "I" selected ahead of
time was printed? Didn't the word recognition software display a result
initially? Or does it offer you a choice of responses?
>
>
>
>> Try speaking the following four sentences one after another and see if
>> your speech recognition software gets them all in the same order:
>>
>> The three words to too and two are homonyms.
>> The three words too to and two are homonyms.
>> The three words two to and two are homonyms.
>> The three words two too and to are homonyms.
>
> The above will not work, because you did not
> include an initial word to differentiate the phrases.

It might not work because there's an "and" in there. Except for that, my
sentence has the same structure as yours.
>
>
>
>
> My versions below _did_ include such words,
> which I capitalized to draw your attention to them.
>
>> the THREE words cent scent sent are homonyms

Good interpretation
>>
>> the FOUR words two too two tutu are homonyms

Not as good. The first or second "two" should have been "to."
>>
>> the TWO words tutu tutu are homonyms

I'm puzzled here. You repeated the same word, didn't you? Did you say
"toto toto" with a perhaps slight pause after the second syllable or did
you say "to to to to" with a slight pause between all four words? Or was
the SR software smart enough to recognize from the TWO that there were
only two words here?
>
>
>
> Think of it this way. If MacSpeech or Dragon hears
> the word "three", it knows that the next word is likely
> to be "words".
>
> When it hears the word "words", it knows that the next
> word is likely to be "cent" - - - NOT "sent" or "scent".
>
> In other words, it gives priority to the word "cent",
> thanks to a document it scanned previously.

Aha, I assume you're saying that Dragon is learning from one or more
previous documents, where "cent" shows up more often than "scent" or "sent".
>
>
> I know it is tough to comprehend, but that is the way
> that the SR app works - - - and the reason why "typical"
> user docs are scanned ahead of time.
>
> Without this particular feature, I would not be able to
> get 100% raw initial accuracy after dictating 600
> highly technical medical words at 150 wpm.
>
> Mark-

But, aside from how word recognition software works, applying English
grammar rules tutu is not a homonym of two, because it is not pronounced
the same -- it has two syllables instead of one.

"Mark Conrad" <> wrote in message
news:251220090551359335%...
> In article <>, andy t

Below was all dictated with WSR and because I said 'tutu' all in one word
with no pauses and in context, it came out fine.

I am wearing a tutu because I want to go down the road and wear it two times
so I do not wear it too much.

Below was all dictated with Dragon and 'tutu' came out fine when dictated in
context when and with no pauses. The only difference between the 2 apps is
that WSR spelt the written word 'two' whereas Dragon has used the numeral
'2'.

I am wearing a tutu because I want to go down the road and wear it 2 times
so I do not wear it too much.



>> The second 2 identical lines below the first two, I have dictated with
>> Dragon. You will notice that Dragon has more of a problem than WSR does
>> in
>> picking out the written form 'two' as opposed to the numeral 2.
>
> That behavior can be changed in Dragon.
>
> I do not know the details, because I have not used
> Dragon for awhile.

I know this. You can open Dragon's 'Vocabulary Editor' and change to
'written word' but it is already hardcoded into WSR so you do not need to do
this.

>> There you go Mark, now you have learned something
>> and that 'tutu' is not a homonym.
>
> It certainly is. Below is a list of two homonyms:
>
> tutu tutu

Um, not what you said. Let me remind you of exactly what you stated.
This is one of your statements from your first post, and I quote;

> "the four words two too two tutu are homonyms"

"No they are not! 'tutu' is not a homonym of 'to, 'too' or 'two'. 'tutu'
is a completely different word and has 2 syllables whereas the others
have only one syllable as Bill said.

>
> Here is another list of two homonyms:
>
> tutu two

Incorrect. 'Tutu' and 'two' are completely different words and not
homonyms. You are mistaken.

If you wanted to be pedantic about it, none of the words you have used in
these examples is true homonyms; they are actually 'homophones'. True
homonyms not only sound the same but are spelt the same as well. I know in
the U.S. it is somewhat different.

Check out the link below: and

http://www.rossde.com/malaprops/homonyms.html

If you had said 'tutu' & 'two two' were homonyms, then you would have been
correct. Once again, check out the link above for this and get your facts
straight first Mark before divulging your bullshit.

Go to the 'T's, look up 'to', 'too' and 'two' (2) and you will see they are
definitely not homonyms of 'tutu'.

Then go down a little bit further and you will see your 'tutu'. The only
homonym you will see beside it, will be 'two' two'. These are two syllable
words that match each other, hence they are the same sounding. You are
confused.


> homonym (noun)
> each of two words having the same pronounciation

You mean, each of the two words have the same pronunciations

'Tutu' and 'two' are 2 different words and are pronounced differently so
they are not homonyms of each other.

I know what a homonym is. You obviously do not!

> list (noun)
> a number of connected items written or printed
> consecutively

Stop trying to change the subject and wriggle out of this. I have already
proved you wrong on the above word's definition. We are now on the subject
of the definition of a homonym and you are wrong in this case :-)

andy t

"Mark Conrad" <> wrote in message
news:271220090756035065%...
> In article <>, Bill Sharpe
> <> wrote:

> >> Do you really believe that "tutu," a two-syllable word
> >> has the same pronunciation as "two," a one-syllable word?
> >
> > Of course not, I never said they did have the same
> > pronunciation.
>
> Well, then, tutu is not a homonym of two.

> Wrong. Easy to disprove your above statement.

No, you are wrong and it is easy to prove you wrong. Look at the link in my
previous post which proves you are 100% wrong.
You say it is easy to disprove the statement. Then prove it! You cannot,
because Bill and I are right!

> You *DO* agree "tutu tutu" is a homonym,
> don't you?

No. Too-too and tutu are homonyms. Look at the link in my last post. I
actually made a mistake in my last post where I said 'two-two' & 'tutu' were
homonyms. I got that wrong. I meant to say that too-too & tutu are
homonyms. Check the link in my last post that proves you are wrong Mark,
which is not surprising as you are most of the time and then you try to
backtrack with your usual bull-****!

> For example, I spoke the following "utterance"
> at 210 wpm

bull-****!

> the THREE words cent scent sent are homonyms

Correct!

> the FOUR words two too two tutu are homonyms

Incorrect! 'Two', 'too' & 'to' are not homonyms of 'tutu'. Fact.

> But, aside from how word recognition software works,
> applying English grammar rules tutu is not a homonym
> of two, because it is not pronounced the same --
> it has two syllables instead of one.

> Please speak the following, count the syllables:

> tutu tutu
> tutu toto
> tutu twotwo
> tutu tootoo
> tutu totwo
> tutu totoo
>...etc., etc.

What? Not relevant!

> ...except when I speak the above it is:

> tutututu
> tututoto
> tututwotwo
> tututootoo
> tututotwo
> tututotoo

What? Utter bull-****.

> We are quibbling over semantics.

No, you are trying to wriggle out of your usual bull-****!

> My present highest speed is 250 wpm which any
> human would have little difficulty understanding.

Here we go! You are dreaming again Mark and talking utter bull-****.

>I have to slow down to 150 wpm in order get !00%
>raw recognition accuracy from either MacSpeech
>or Dragon, in 3 out of 4 tries on technical medical
>dictation of 600 words.

You mean as you have above. :-) Look at your 100% mistake above. :-)

You would be very lucky to even get 150 words per minute without any
mistakes and 100% accuracy.

>Can't get this level of accuracy with WSR (Vista speech)

I can.

>There is a name for this phenomenon, but I
>forget what that name is;

Yes, in your case it is called bull-****!

>It does not bother humans, however; we have no
>trouble understanding speech at even 300 wpm.

Even more bull-****!

Your answer to everything is think of a number then treble it! You talk
utter nonsense!

andy t

See inline comments

Bill

Mark Conrad wrote:
> In article <>, Bill Sharpe
> <> wrote:
>
>>>> Do you really believe that "tutu," a two-syllable word
>>>> has the same pronunciation as "two," a one-syllable word?
>>> Of course not, I never said they did have the same
>>> pronunciation.
>> Well, then, tutu is not a homonym of two.
>
> Wrong. Easy to disprove your above statement.
>
> If you hear me say something that sounds like:
>
> "to to to to"
>
> ...then is the homonym "tutu" anywhere in there?

If I hear you say the above, I don't know what you mean -- perhaps you
are stuttering. It could be any combination of to, 2, two, too, and,
yes, tutu. That doesn't make tutu a homonym of any of the other items or
of itself.
>
>
> You *DO* agree "tutu tutu" is a homonym,
> don't you?

Of course not. That's what I've been saying all along. Two words are
homonyms if they are pronounced or spelled the same way but have
different meanings. tutu tutu is the same word repeated twice; the
second tutu doesn't have a different meaning than the first.
>
>
> I repeat my question, is the homonym "tutu"
> anywhere inside the phrase that sounds like:
>
> "to to to to"
>
See above

Mark Conrad wrote:
> Okay Bill, here is my hour-later post regarding
> the subject of scanning documents to improve
> accuracy.
>
> Seems I was wrong about a basic concept.
>
> The "trip word" (two, three, four, etc.)
> does NOT have to be spoken a few words before
> the list of homonyms, it can be spoken a great
> distance before the list of homonyms.
>
> Sorry for the confusion I introduced initially
> regarding this subject.
>
> I am still learning, as are the rest of us.
>
>
> The important thing is that the document scanning
> feature HAS to work. It does in both MacSpeech
> and Dragon.
>
> I can't get it to work in WSR. (Vista speech)
>
> Therefore the following can't be done reliably and
> consistantly in the Vista speech app.
>
> "four homonyms are to too two tutu"
>
>
> ...spoken rapidly without pauses as in:
>
> "fourhomonymsaretototototo"
>
> Because of this apparent flaw in WSR...
>
> ...namely its inability to scan documents like
> both Dragon and MacSpeech can, it is liable to
> yield as many as 30 text mistakes is a 600 word
> medical dictation session, instead of the zero
> mistakes that both Dragon and MacSpeech yield.
>
> That document scanning feature is absolutely
> essential to radically reduce text mistakes
>
> Mark-

I will certainly agree with you, Mark, that Dragon is a much better SR
program than WSR. Of course Dragon is a stand-alone product that must be
purchased separately. WSR is packaged with recent versions of Windows
and doesn't cost extra; I do not use it on a regular basis, but it seems
to work fine for occasional dictation, speaking at a somewhat slower
pace than normal.

Did Dragon give you zero mistakes the first time you dictated your
medical information? From what you've indicated, Dragon had to learn
from either previously scanned documents or practice speech by you.

Bill

andy t wrote:
> Bill, I suggest you ignore this individual. He is so thick-skinned, you
> will be having this conversation with him forever because he will never
> accept that he is wrong.
>
> He will not take advice and except he is wrong even when the proof is
> put in front of him. He lives in a world of his own, so my advice would
> be to leave him be.
>
> He is not willing to learn and he is set in his ways and only believes
> what he wants to believe.
>
> He thinks he is always right but most of the time he is far from it.
>
> He is in my opinion, a lost cause and you will be wasting your time
> trying to teach him anything. I am afraid there is no hope for this sad
> little old man. :-)
>
> andy t
>
> "andy t" <> wrote in message
> news:...
>>
>>
>> "Mark Conrad" <> wrote in message
>> news:271220091509469725%...
>>>
>>> Okay Bill, here is my hour-later post regarding
>>> the subject of scanning documents to improve
>>> accuracy.
>>>
>>> Seems I was wrong about a basic concept.
>>
>> You are wrong about a lot of things as usual.
>>
>>> I am still learning, as are the rest of us.
>>
>> How long did you say you have been learning speech recognition? Seems
>> like
>> you may need another 20 years.
>>
>>> "four homonyms are to too two tutu"
>>
>> Incorrect. My oh my , you really do have a lot to learn!
>>
>> 'To', 'too' and 'two' are not homonyms of 'tutu'. 'Too-too' and
>> 'tutu' are
>> both homonyms. Look at the link below. It's so easy to prove you
>> wrong :-)
>>
>> http://www.rossde.com/malaprops/homonyms.html
>>
Thanks for your suggestions Andy. The only point on which I can agree
with Mark is his comment that Dragon is much better than Windows Speech
Recognition. Based on the cost of the two programs I am not surprised at
this result. He doesn't seem to know much about homonyms, though.

Bill