Re: Speech Recognition (SR) - Just How Useful Is It ?

I got into it for a while. It's pretty cool only if you are alone. It
seems to irritate the crap out
of others around.

"Mark Conrad" <> wrote in message
news:020320100702068280%...
>
> Hopefully this will be a constructive upbeat productive
> thread, unless the usual trolls pollute this thread.
>
> (SR = "Speech Recognition" to save space)
>
> Just how useful is SR, to Windows users?
>
> That of course depends on the user.
>
> One could say that SR is 100% useless to 99% of the
> people in this NG, which would be a compelling argument,
> difficult to dispute.
>
> However we Windows users get SR free, so it would be
> a shame to waste it, *IF* there is any reason at all
> for us to use SR.
>
> So, is there a reason?
>
> Yes, a "free form" blog is one reason.
>
> If you have anything at all worth "talking" to another person,
> then that is a candidate for doing faster and easier with SR.
>
> Of course if you have NOTHING worth talking about...
>
>
> So it all boils down to _your_ "worth" to other people,
> as perceived by those people themselves.
>
>
> To close this post, I have one more thing worth mentioning,
> namely that our thoughts are fleeting, Great Thoughts often
> "get lost in the shuffle".
>
> It is only when we commit those thoughts to permanent text
> that those Great Thoughts can do any good for other people,
> or even for ourselves.
>
> SR is an excellent way to "capture" your Great Thoughts;
> I carry around a 3 ounce digital recorder just for this
> purpose - - - and yes, you have guessed right, my real
> name is Bill Gates, and I owe my fortune to that tiny
> digital recorder, which is one important difference
> between me and you ordinary mortals.
>
> ...because I CAPTURED my great thoughts, while you let
> your great thoughts slip through you fingers.
>
> Mark-

Are you a doctor or professor of some sort?

I have also tried it but with little success. Do you need to spend a lot of
time with it?


"Mark Conrad" <> wrote in message
news:050320100103038987%...
> In article <9205347E-3303-4846-BC04->, Jimmy
> <> wrote:
>
>> I got into it for a while. It's pretty cool only if you are alone. It
>> seems to irritate the crap out of others around.
>
> Yeah, kinda like cell-phones.
>
> I like to get some dork in my supermarket who is
> jabbering on his/her phone, then start my own dictation
> into an Olympus DS-5000 digital recorder - - -
> which is hidden in my shirt pocket.
>
> I actually had some woman walk around me furtively,
> looking for my "cellphone".
>
> When she could not spot one, she tippy-toed away from
> the guy she thought was talking to himself. <g>
>
> My DS-5000 is designed for speech recognition use,
> darn thing costs $600.
>
> It handles medical crap quite well, words like
> "perioperative transesophageal echocardiography" etc
> without needing to be tethered to a PC.
>
>
> Lotsa people play with SR for a while, say "that's nice", then
> drop it entirely because they can't figure any use for it.
>
> The usual "turn off" is that they just get annoyed with correcting
> text mistakes the stupid software makes.
>
>
> Doctors have figured a way out of that nasty chore, they just
> make someone on their staff correct the mistakes, by comparing
> the original audio against the text.
>
>
>
> That way the doc' can roller-skate between twice as many
> exam' rooms, giving each patient the full 60 seconds time they
> are entitled to.
>
> Mark-

The spammer Marty Martin Markoe and his buddies use many aliases
for advertising their commercial website and for attacking their
adversaries, for example...

http://www.knowbrainer.com/
"After tracing this members profile we are embarrassed to find out
that we now know that Mike Tavia is another one of approximately 20
KnowBrainer forum aliases for Martin Markoe of emicrophones who was
banned from this forum for personal attacks."

Here is an example of Marty Martin Markoe & Company masquerading
as a lurker on USENET.
http://groups.google.com/group/comp....6f446a24e298e?
In fact... that fraud was his only post to USENET and his NNTP
posting host (assigned to optonline.net) matched his buddy Michael
Mendick.

eMicrophones MyMSSpeech stalks, slanders, threatens others, and
publicly publishes information from private investigators such as
Kelmar & Associates, violating the privacy of innocent people who
have nothing to do with fighting eMicrophones MyMSSpeech spam.

Following are a few examples that illustrate the annoyance and
frustration eMicrophones MyMSSpeech spam from Martin Markoe causes
speech users seeking help.

http://groups.google.com/group/comp....a2b3f9dd6a0f0?
"Man you really are full of yourself [Martin]! Just leave me
alone... go away and let me get USEFUL information from someone
who will be honest with me instead of evasive, devisive and
insulting!"

http://groups.google.com/group/comp....cc9fa6c340ffc?
http://groups.google.com/group/comp....a2b3f9dd6a0f0?
http://groups.google.com/group/comp....6b8b7563712e9?
http://groups.google.com/group/comp....4c0e38032578f?
http://groups.google.com/group/comp....7c2d47a1d5994?
http://groups.google.com/group/comp....3180162c1d199?
http://groups.google.com/group/comp....cc5fd879272fd?
http://groups.google.com/group/comp....362b69d93a460?
http://groups.google.com/group/comp....5a5e972b1cc1c?

See also:
eMicrophones, MyMSSpeech
billyrichbroker yahoo.com
M.M. <twatface41 yahoo.com>
MartyMarkoeeMicrophonesInc discussions.microsoft.com
mmarkoe gmail.com
mmarkoe yahoo.com
Martin Markoe <martin emicrophones.com>
Martin Markoe <martin speechcontrol.com>
Martin Markoe <mmarkoe optonline.net>
Michael Mendick <michael emicrophones.com>
NEWSGROUP DESTROYER <twatface41 yahoo.com>
wbonneau gmail.com
wwwMyMSSpeechcom discussions.microsoft.com


mmarkoe <mmarkoe gmail.com> wrote:

> Path: news.astraweb.com!border1.newsrouter.astraweb.com! indigo.octanews.net!news-out.octanews.net!mauve.octanews.net!nx01.iad01.new shosting.com!newshosting.com!novia!news-out.readnews.com!news-xxxfer.readnews.com!postnews.google.com!t41g2000yq t.googlegroups.com!not-for-mail
> From: mmarkoe <mmarkoe gmail.com>
> Newsgroups: microsoft.public.windows.vista.general
> Subject: Re: Speech Recognition (SR) - Just How Useful Is It ?
> Date: Sat, 6 Mar 2010 04:31:56 -0800 (PST)
> Organization: http://groups.google.com
> Lines: 29
> Message-ID: <560eca3b-7d42-4c7e-9ac1-766e06440bbb t41g2000yqt.googlegroups.com>
> References: <020320100702068280%aeiou mostly.invalid> <9205347E-3303-4846-BC04-4A6C233DD5F6 microsoft.com> <050320100103038987%aeiou mostly.invalid> <uSKCoiQvKHA.6140 TK2MSFTNGP05.phx.gbl>
> NNTP-Posting-Host: 67.87.76.112
> Mime-Version: 1.0
> Content-Type: text/plain; charset=ISO-8859-1
> Content-Transfer-Encoding: quoted-printable
> X-Trace: posting.google.com 1267878716 17226 127.0.0.1 (6 Mar 2010 12:31:56 GMT)
> X-Complaints-To: groups-abuse google.com
> NNTP-Posting-Date: Sat, 6 Mar 2010 12:31:56 +0000 (UTC)
> Complaints-To: groups-abuse google.com
> Injection-Info: t41g2000yqt.googlegroups.com; posting-host=67.87.76.112; posting-account=wCRjrwoAAAD8nAAOusCMQoFPZqKsecHb
> User-Agent: G2/1.0
> X-HTTP-UserAgent: Mozilla/4.0 (compatible; MSIE 7.0; Windows NT 6.0; SLCC1; .NET CLR 2.0.50727; .NET CLR 1.1.4322; .NET CLR 3.5.21022; InfoPath.2; .NET CLR 3.5.30729; .NET CLR 3.0.30729),gzip(gfe),gzip(gfe)
>
> On Mar 6, 3:40ÿam, "Rod" <rodarn... hotmail.com> wrote:
>
>> I have also tried it but with little success. ÿDo you need to spend a l
> ot of
>> time with it?
>
> Speech recognition software like Dragon and WSR really work and can
> work well. A good place to start is to read the short article,
> "Getting Started with Windows Speech Recognition," at:
> http://www.mymsspeech.com/download/WSRGetStarted.pdf
>
> In a nutshell, you need good hardware. This includes a good noise
> canceling microphone and a sound card that is well shielded and does
> not introduce electronic noise from within the computer. It is also
> important to understand how it works and how you must adapt slightly
> to how it works. Speech recognition software first looks for the
> sounds of the words you make. By enunciating each word clearly you
> help it guess which word you said. Next it compares each word to the
> words around it for context clues. This is how it knows which word to
> (two or too) to use as in the example, "Two boys went to see a doctor
> because they ate too much food." By speaking in phrases and
> enunciating each word, you assist the system to figure out what you
> actually say and mean.
>
> We routinely get 99% accuracy and have since beginning to use speech
> recognition software back in 1994.
>
> Marty Markoe, eMicrophones, Inc.
> http://www.MyMSSpeech.com
>
>

"Rod" <rodarnold hotmail.com> wrote:

> I have also tried [speech recognition] but with little success.
> Do you need to spend a lot of time with it?

Yes. You must train your voice until the stupid computer can
recognize your speech. One way I did so was to buy a digital voice
recorder (Olympus has made good ones) and take notes. When you
record a note, press the playback button and listen. Repeat that
until it sounds good to you. I spent over a year doing that before
being able to cope with speech recognition. BTW, a digital
recorder is cheap and can be useful for other purposes too.

Beware of malicious scheming spammers like Marty Martin Markoe who
take advantage of aspiring speech recognition users and would
entice you to buy their expensive hardware. Their racket goes like
this... They tell you to buy their hardware. It does not help.
They tell you to buy more of their hardware. It still does not
help. Then they throw up their hands and say "well, it helps
everyone else". The implication is that your voice sucks. They do
not have to say it, but you realize it, so all you can do is sulk
and walk away.

Your voice has the far greatest influence over whether speech
recognition will work for you. I have been using the same
inexpensive USB microphone for years. It works great. I recently
purchased another inexpensive USB microphone and it works even
better, but I usually use the old one because it has been modified
to work with my computer setup.

Good luck and have fun.
--




















> From: "Rod" <rodarnold hotmail.com>
> References: <020320100702068280%aeiou mostly.invalid> <9205347E-3303-4846-BC04-4A6C233DD5F6 microsoft.com> <050320100103038987%aeiou mostly.invalid>
> In-Reply-To: <050320100103038987%aeiou mostly.invalid>
> Subject: Re: Speech Recognition (SR) - Just How Useful Is It ?
> Date: Sat, 6 Mar 2010 08:40:09 -0000
> Lines: 1
> MIME-Version: 1.0
> Content-Type: text/plain; format=flowed; charset="iso-8859-1"; reply-type=original
> Content-Transfer-Encoding: 7bit
> X-Priority: 3
> X-MSMail-Priority: Normal
> Importance: Normal
> X-Newsreader: Microsoft Windows Live Mail 14.0.8089.726
> X-MimeOLE: Produced By Microsoft MimeOLE V14.0.8089.726
> Message-ID: <uSKCoiQvKHA.6140 TK2MSFTNGP05.phx.gbl>
> Newsgroups: microsoft.public.windows.vista.general
> NNTP-Posting-Host: 79-74-132-146.dynamic.dsl.as9105.com 79.74.132.146
> Path: news.astraweb.com!border5.newsrouter.astraweb.com! news-xfer.nntp.sonic.net!news.alt.net!msrtrans!TK2MSFTF EEDS02.phx.gbl!TK2MSFTNGP01.phx.gbl!TK2MSFTNGP05.p hx.gbl
>

Mark Conrad <> wrote:

> Almost always it is because I mispronounce a word, such as
> "transesophageal".
>
> trans-s-o-fay-geal
>
> Actual doctors are amused by my butchering of medical
> pronunciation, but heck, it works for me.<g>

transesophageal
antidisestablishmentarianism

I have no trouble with long words. The speech recognition is
probably more likely to get a long word right, if it is in the
vocabulary.

Transesophageal
transesophageal
transesophageal
transesophageal
transesophageal
transesophageal
transesophageal
transesophageal
transesophageal
....

The real problem is with trying to get the computer to understand words
in normal speech.

Mark Conrad <> wrote:

> In article <4b961512$0$13556$>, John Doe
> <> wrote:
>
>> Mark Conrad <> wrote:
>>
>> > Almost always it is because I mispronounce a word, such as
>> > "transesophageal".
>> >
>> > trans-s-o-fay-geal
>> >
>> > Actual doctors are amused by my butchering of medical
>> > pronunciation, but heck, it works for me.<g>
>>
>> transesophageal
>> antidisestablishmentarianism
>>
>> I have no trouble with long words. The speech recognition is
>> probably more likely to get a long word right, if it is in the
>> vocabulary.
>>
>> Transesophageal
>> transesophageal
>> transesophageal
>> transesophageal
>> transesophageal
>> transesophageal
>> transesophageal
>> transesophageal
>> transesophageal
>> ...
>>
>> The real problem is with trying to get the computer to understand
>> words in normal speech.

> Doc' scanning, doc' review, whatever one cares to call it,
> handles those short one-syllable words based on their
> association with prior words in a previously scanned doc'.
>
> Best example I can think of right now is:
>
> The homonyms are cent sent scent
>
> ...spoken normally like this:
>
> Thehomonymsarecentsentscent

That cannot be understood without defining the thing. There is no way
to know which order you want the words.

I just do correcting as needed. That was probably the most difficult
thing to get used to, always being ready to correct the computer. I
guess one of the initial hangups is having to interact so closely
with the stupid thing... yuck.

And then there are times when I know that my words were slurred
horribly and the stupid computer gets them right anyway...

A one or two syllable word spoken with no context can be very tough
here.

Anyway... I am long past the point of being overly annoyed that the
stupid computer from time to time misrecognizes stuff I say. I guess
you learn the percentages, when it is more likely to make mistakes,
and you are ready to make the correction. At least the decent stuff
like DNS is reasonably easy to predict, especially with experience.
My early days with ViaVoice, I could not even guess when it was going
to have trouble recognizing my speech and when it might properly
recognize it, it was unpredictable.

On 3/9/2010 12:34 PM, Mark Conrad wrote:
> In article<0075d1af$0$2924$> , John Doe
> <> wrote:
>
>> That cannot be understood without defining the thing. There is
>> no way to know which order you want the words.
>
> Actually I just snapped out that (poor) example.
>
> A better example would be:
>
> Alphabetical order of homonyms are sent scent sent
>
>
>
>> That cannot be understood without defining the thing.
>
> True, the previously scanned document "defines" the
> order to be: sent scent sent.
>
which, of course, is not alphabetical. And sent and sent are the same
word, not homonyms.

Of course you'd have to have a defining document for the following
absurd phrase to get it correct:
Willy sent Millicent a silly scent with his strange flowers.

Bill