Speech Feedback to Recogniser

  • Thread starter Thread starter HardySpicer
  • Start date Start date
H

HardySpicer

I am writing my own recogniser and synthesis software in VB .net.
However, every time I get the syntheser to speek something the mic
picks it up and thinks it is a command! It is quite bizzar - it sits
there talking to itself pure jibberish! I use Microsoft Speech the
latest SDK. I tried turing off teh recogniser of course but it doesn't
seem to work. If I turn down the volume of the speakers it's ok of
course but I want to hear the speech. Any ideas?

Hardy
 
HardySpicer said:
I am writing my own recogniser and synthesis software in VB .net.
However, every time I get the syntheser to speek something the mic
picks it up and thinks it is a command! It is quite bizzar - it
sits there talking to itself pure jibberish! I use Microsoft
Speech the latest SDK. I tried turing off teh recogniser of course
but it doesn't seem to work. If I turn down the volume of the
speakers it's ok of course but I want to hear the speech. Any
ideas?

Hardy

The speech users group is for users. You can try the research group.
Good luck.
 
John said:
The speech users group is for users. You can try the research group.
Good luck.

Speech recognition users used to have good luck getting answers to
questions on this user group until you, Mark Orrin Bender aka John
Doe, killed off this group because you think you are an Internet
policeman.
 
HardySpicer said:
I am writing my own recogniser and synthesis software in VB .net.
However, every time I get the syntheser to speek something the mic
picks it up and thinks it is a command! It is quite bizzar - it sits
there talking to itself pure jibberish! I use Microsoft Speech the
latest SDK. I tried turing off teh recogniser of course but it doesn't
seem to work. If I turn down the volume of the speakers it's ok of
course but I want to hear the speech. Any ideas?


Hardy:

If you are writing in VB.NET, why are you sending to
microsoft.public.dotnet.languages.vc, which is a Visual C++ group?

As to your question, have you thought of headphones :-) ?
 
eMicrophones Inc Martin Markoe spams USENET, stalks and slanders
other authors, and violates the privacy of innocent people who have
nothing to do with eMicrophones Inc spam or its malicious practices.
By itself, eMicrophones Inc willingness to push microphones on
disabled users whose problems have nothing to do with a microphone
provides insight into their malicious characters.

Anyone can express his (or her) opinion here on USENET, but
advertising is not allowed except in groups specifically labeled for
selling things like *forsale* and *marketplace*.

Following are a few quotes to help illustrate the annoyance and
frustration eMicrophones, Inc spam from Martin Markoe causes users
seeking help in the speech recognition group and why advertising is
prohibited on USENET with explicitly labeled exceptions.

http://groups.google.com/group/comp.speech.users/msg/3becc9fa6c340ffc?

You are not a technologist. You sell microphones.

Marty, how come whenever someone disagrees with you, you get into this
high school urinating contest mode? Actually I can't find your
credentials anywhere either, except as a guy selling something.

http://groups.google.com/group/comp.speech.users/msg/473a2b3f9dd6a0f0?

Man you really are full of yourself! Just leave me alone. I asked you
a question. Silly me, I wanted to know what I was buying before I
bought it. How absurd! I then get email after email after email from
you telling me that you won't answer my question and trying to justify
it. Lets cut the crap. You won't answer my question. I won't buy from
you. So now 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.speech.users/msg/d256b8b7563712e9?

More importantly I am not here touting my business. A position which
makes anything you say suspect, doesn't it.

I have been watching this group for a long time and you have managed
to turn it into the eMicrophones, Inc. mail list. Almost no one with
expertise posts here any longer.

http://groups.google.com/group/comp.speech.users/msg/6014c0e38032578f?

Is anyone else beyond annoyed with the two pathetically desperate
"small business owners" that have driven this group into the ground?

I subscribed 6 months ago hoping to learn something about the
"state-of-the-art" in voice recognition from the user's perspective.

All I've managed to learn is that most users are apparently being
ripped off by over-charging slimy weaseloid microphone hawkers.

Martin Markoe and Mr. "speaktoyourpc.com", I beg you to consider the
number of potential customers you're pissing off before you further
demonstrate your witlessness by posting again. Your childish
King-of-the-Group games should've stopped a loooong time ago.

http://groups.google.com/group/comp.speech.users/msg/3b57c2d47a1d5994?

Well said...Refreshing.

[That in response to the prior message.]

http://groups.google.com/group/comp.speech.users/msg/ce73180162c1d199?

.... dealing with you on the newsgroups is not an easy thing and most
always turns into some type of flaming war.

.... you do get into fights with many regarding microphones and you do
wave your website around more often than necessary and you don't
always tell the truth about competitive products.

I didn't respond because I simply did not want to get into another
fight which is the reason my posts have been very light in the last
few months.

Here are my qualifications. Certified and trained for Philips
SpeechMagic, IBM ViaVoice, IBM Medspeak, Kurzweil VoiceMed, Dragon
NaturallySpeaking Enterprise & L&H VoiceXpress... So please, don't
call me a "Charlatan" because it is simply not true.

And believe me when I tell you that some users actually get better
recognition from $20 headsets than with $100 headsets.

I see vendors posting all the time and as long as you are not a member
of the newsgroup, there aren't constant one-up battles.

http://groups.google.com/group/comp.speech.users/msg/ba6cc5fd879272fd?

After getting caught with his pants down regarding his so called
privacy policy, Martin continues to make a complete ass out of
himself.

With this pattern of behaviour, the next step is usually
stalking--email stalking (... to menacingly spread deceit)

[Martin Markoe does that.]

http://groups.google.com/group/comp.speech.users/msg/84f362b69d93a460?

.... you don't see the authors in all the newsgroups in a constant
battle. Only with you Marty.

I have seen posts from you that link to as many as 8 direct links to
your website (all in one post) and your signatures fall within 5 or 10
lines long. In my book, that's called SPAMMING.

.... about 2 weeks ago, you claimed to have sold thousands of GN MPA
Satellites. I know for a fact that is not true.

Guess what Marty.... I have a life. I don't have time to live on the
newsgroups as you do.

This is the 2nd time you have mentioned me being moderated on user
groups. Last time, you apologized about confusing me with someone
else. YOU ARE OUT OF CONTROL MARTY MARKOE.

Marty, your constant abuse on these newsgroups... must come to an end.

http://groups.google.com/group/comp.speech.users/msg/5585a5e972b1cc1c?

I give up. I've dealt with delusional people like you before, and it
leads to nowhere... you continue attacking people who are FAR more
experienced and qualified than you, not to mention professional.

See also:
"Martin Markoe" <martin emicrophones.com>
.... <martin speechcontrol.com>
.... <mmarkoe optonline.net>
"Michael Mendick" <michael emicrophones.com>
billyrichbroker yahoo.com


************************************************************



Path: newssvr21.news.prodigy.net!newsdbm04.news.prodigy.net!newsdst01.news.prodigy.net!prodigy.com!newscon04.news.prodigy.net!prodigy.net!news.astraweb.com!router1.astraweb.com!news.glorb.com!postnews.google.com!a28g2000hsc.googlegroups.com!not-for-mail
From: Martin Markoe <mmarkoe optonline.net>
Newsgroups: comp.speech.users, microsoft.public.dotnet.languages.vc
Subject: Re: Speech Feedback to Recogniser
Date: Sun, 18 Nov 2007 18:23:25 -0800 (PST)
Organization: http://groups.google.com
Lines: 15
Message-ID: <fc9f6667-9ed7-457e-bd5a-f940ea930eac a28g2000hsc.googlegroups.com>
References: <4255de3c-261e-414c-b345-447b689086ff s36g2000prg.googlegroups.com> <QM20j.18738$4V6.6893 newssvr14.news.prodigy.net>
NNTP-Posting-Host: 69.119.117.84
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
X-Trace: posting.google.com 1195439006 5896 127.0.0.1 (19 Nov 2007 02:23:26 GMT)
X-Complaints-To: groups-abuse google.com
NNTP-Posting-Date: Mon, 19 Nov 2007 02:23:26 +0000 (UTC)
Complaints-To: groups-abuse google.com
Injection-Info: a28g2000hsc.googlegroups.com; posting-host=69.119.117.84; posting-account=tyYSzQoAAAABZ6kBXCYp_Nb3X6VPblvh
User-Agent: G2/1.0
X-HTTP-UserAgent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; .NET CLR 1.1.4322; .NET CLR 2.0.50727),gzip(gfe),gzip(gfe)
Content-Disposition: inline
Xref: prodigy.net comp.speech.users:14843 microsoft.public.dotnet.languages.vc:65806

John said:
The speech users group is for users. You can try the research group.
Good luck.

Speech recognition users used to have good luck getting answers to
questions on this user group until you, Mark Orrin Bender aka John
Doe, killed off this group because you think you are an Internet
policeman.
 
Since you control both the recognizer and the synthesizer and you know when
the synthesizer is speaking you could turn off the recognizer or ignore the
SR output when the synthesizer is speaking.

In future releases we are looking at incorporating audio echo cancellation
into the signal path which should eliminate this kind of feedback.
 
I had the EXACT same problem about a month ago! I created a speech
recognizing program that confirmed via repeating the command ala
text-to-speech, only to discover that the recognizer recognized this
text-to-speech, and it went into an infinite loop of the command being
spoken and acted on!

I fixed this by just turning off speech recognition during text-to-speech
talking. This is easy to do via the SpeakStarted and SpeakCompleted events.

There is another way though, if your application can work within this
limitation. Just make sure nothing the text-to-speech says is not something
that can be recognized. That is, nothing spoken via TTS is something within
the grammars of the speech recognizer.
 
In future releases we are looking at incorporating audio echo cancellation
into the signal path which should eliminate this kind of feedback.

Really? That would be pretty dang cool!

If this is done, I'd like to suggest the ability to use a user-selectable
sound source as the basis of cancellation. Thus, a person could set up a
cancelling microphone on, say, an external 'noise maker' and have it
filtered out.

For example, a cancelling microphone could be placed near a TV in the room
and then the TV would no longer be listened to for SR!

I'm in no way familiar with the problems and technology involved, so I could
be asking for the impossible here. But thought I'd mention it anyway...
 
Back
Top