System.Speech questions

Hi Team,

I have some questions speech recognition through System.Speech.

1.For deployment on an XP box, System.Speech seems to use the existing SAPI. 
What issues are there using System.Speech with SAPI 5.1 instead of 5.3?
2.How do we maintain good performance, particularly when generating dynamic 
grammars?

3.Is there a performance or optimization benefit to training the engine?

4.How do we localize applications using System.Speech to take advantage of 
the eight languages supported out of the box?

5.SAPI 5.1 seems to want to reset the microphone settings (particularly 
volume) every time the engine is started; however, these settings are never 
right (for example, our applications are used in helicopters while in flight 
-- this requires different settings that the average user).Does System.Speech 
provide controls to embed the microphone control directly into our 
applications, so that users can adjust the speech settings without needing 
direct access to the control panel? What permissions issues arise from trying 
to set the microphone volume programmatically in a constrained-permissions 
environment?
a.Same question but for controls for speech accuracy/responsiveness settings.

Thanks in advance,

1.) You can use an application on Windows XP that uses the System.Speech 
namespace but you need to watch out for features that System.Speech has 
support for SAPI 5.3, if you use those features in your application they 
won't work in Windows XP SAPI 5.1

2.) What level of performance are you talking about? The amount of words the 
recognizer will recognize or hardward type performance?

3.) The benefit to training is recognition more than benefit, by doing 
training the recoginzer will be able to pick up words for the specific user, 
such as if you have an accent it might be hard for the recognizer to 
understand certain words, by doing training it will be easier for the it to 
pick those words up.

4.) The 8 languages which you refer to aren't really System.Speech features 
but SAPI 5.3 feature. System.Speech was really geared to work with 5.3, while 
maintaining compataibility with 5.1. (It doesn't extend any features to 5.1, 
you can only use features which 5.1 already has.)

5.) I don't think you can change the microphone volume using the 
System.Speech namespace. I haven't done it my self but I will check into it.

-- 
Michael D. Dunn

Magenic Technologies
www.magenic.com
blog: http://gotspeech.net/blogs/michaeldunn


"Vijay Pathak" wrote:

> Hi Team,
> 
> I have some questions speech recognition through System.Speech.
> 
> 1.For deployment on an XP box, System.Speech seems to use the existing
SAPI. 
> What issues are there using System.Speech with SAPI 5.1 instead of 5.3?
> 2.How do we maintain good performance, particularly when generating dynamic

> grammars?
> 
> 3.Is there a performance or optimization benefit to training the engine?
> 
> 4.How do we localize applications using System.Speech to take advantage of

> the eight languages supported out of the box?
> 
> 5.SAPI 5.1 seems to want to reset the microphone settings (particularly 
> volume) every time the engine is started; however, these settings are never

> right (for example, our applications are used in helicopters while in
flight 
> -- this requires different settings that the average user).Does
System.Speech 
> provide controls to embed the microphone control directly into our 
> applications, so that users can adjust the speech settings without needing

> direct access to the control panel? What permissions issues arise from
trying 
> to set the microphone volume programmatically in a constrained-permissions

> environment?
> a.Same question but for controls for speech accuracy/responsiveness
settings.
> 
> Thanks in advance,

Thanks for the post, Michael.

Regarding the second question, here is more information:

"We're using the speech engine for command and control in a data collection

app. However, since the controls presented are based on a user-configurable 
hierarchy, we don't know what grammars we'll need until run-time. So we 
dynamically generate grammars as they're required. In this type of scenario, 
is there a recommend technique to improve the performance of the system when 
switching between custom grammars? "

Hope this helps.

"Michael Dunn" wrote:

> 1.) You can use an application on Windows XP that uses the System.Speech 
> namespace but you need to watch out for features that System.Speech has 
> support for SAPI 5.3, if you use those features in your application they 
> won't work in Windows XP SAPI 5.1
> 
> 2.) What level of performance are you talking about? The amount of words
the 
> recognizer will recognize or hardward type performance?
> 
> 3.) The benefit to training is recognition more than benefit, by doing 
> training the recoginzer will be able to pick up words for the specific
user, 
> such as if you have an accent it might be hard for the recognizer to 
> understand certain words, by doing training it will be easier for the it to

> pick those words up.
> 
> 4.) The 8 languages which you refer to aren't really System.Speech features

> but SAPI 5.3 feature. System.Speech was really geared to work with 5.3,
while 
> maintaining compataibility with 5.1. (It doesn't extend any features to
5.1, 
> you can only use features which 5.1 already has.)
> 
> 5.) I don't think you can change the microphone volume using the 
> System.Speech namespace. I haven't done it my self but I will check into
it.
> 
> -- 
> Michael D. Dunn
> 
> Magenic Technologies
> www.magenic.com
> blog: http://gotspeech.net/blogs/michaeldunn
> 
> 
> "Vijay Pathak" wrote:
> 
> > Hi Team,
> > 
> > I have some questions speech recognition through System.Speech.
> > 
> > 1.For deployment on an XP box, System.Speech seems to use the existing
SAPI. 
> > What issues are there using System.Speech with SAPI 5.1 instead of
5.3?
> > 2.How do we maintain good performance, particularly when generating
dynamic 
> > grammars?
> > 
> > 3.Is there a performance or optimization benefit to training the
engine?
> > 
> > 4.How do we localize applications using System.Speech to take
advantage of 
> > the eight languages supported out of the box?
> > 
> > 5.SAPI 5.1 seems to want to reset the microphone settings
(particularly 
> > volume) every time the engine is started; however, these settings are
never 
> > right (for example, our applications are used in helicopters while in
flight 
> > -- this requires different settings that the average user).Does
System.Speech 
> > provide controls to embed the microphone control directly into our 
> > applications, so that users can adjust the speech settings without
needing 
> > direct access to the control panel? What permissions issues arise from
trying 
> > to set the microphone volume programmatically in a
constrained-permissions 
> > environment?
> > a.Same question but for controls for speech accuracy/responsiveness
settings.
> > 
> > Thanks in advance,

Check this post out: 
http://gotspeech.net/blogs/michaeldunn/archive/2007/02/28/vista-sapi-5-3-speech-
recognition.aspx

Look at the SRGS and the Grammar Builder sections, either one of those 
should be able to help you out a bit. Performance is really going to depend 
how well your grammar is built.
-- 
Michael D. Dunn

Magenic Technologies
www.magenic.com
blog: http://gotspeech.net/blogs/michaeldunn


"Vijay Pathak" wrote:

> Thanks for the post, Michael.
> 
> Regarding the second question, here is more information:
> 
> "We're using the speech engine for command and control in a data
collection 
> app. However, since the controls presented are based on a user-configurable

> hierarchy, we don't know what grammars we'll need until run-time. So we 
> dynamically generate grammars as they're required. In this type of
scenario, 
> is there a recommend technique to improve the performance of the system
when 
> switching between custom grammars? "
> 
> Hope this helps.
> 
> "Michael Dunn" wrote:
> 
> > 1.) You can use an application on Windows XP that uses the
System.Speech 
> > namespace but you need to watch out for features that System.Speech
has 
> > support for SAPI 5.3, if you use those features in your application
they 
> > won't work in Windows XP SAPI 5.1
> > 
> > 2.) What level of performance are you talking about? The amount of
words the 
> > recognizer will recognize or hardward type performance?
> > 
> > 3.) The benefit to training is recognition more than benefit, by doing

> > training the recoginzer will be able to pick up words for the specific
user, 
> > such as if you have an accent it might be hard for the recognizer to 
> > understand certain words, by doing training it will be easier for the
it to 
> > pick those words up.
> > 
> > 4.) The 8 languages which you refer to aren't really System.Speech
features 
> > but SAPI 5.3 feature. System.Speech was really geared to work with
5.3, while 
> > maintaining compataibility with 5.1. (It doesn't extend any features
to 5.1, 
> > you can only use features which 5.1 already has.)
> > 
> > 5.) I don't think you can change the microphone volume using the 
> > System.Speech namespace. I haven't done it my self but I will check
into it.
> > 
> > -- 
> > Michael D. Dunn
> > 
> > Magenic Technologies
> > www.magenic.com
> > blog: http://gotspeech.net/blogs/michaeldunn
> > 
> > 
> > "Vijay Pathak" wrote:
> > 
> > > Hi Team,
> > > 
> > > I have some questions speech recognition through System.Speech.
> > > 
> > > 1.For deployment on an XP box, System.Speech seems to use the
existing SAPI. 
> > > What issues are there using System.Speech with SAPI 5.1 instead
of 5.3?
> > > 2.How do we maintain good performance, particularly when
generating dynamic 
> > > grammars?
> > > 
> > > 3.Is there a performance or optimization benefit to training the
engine?
> > > 
> > > 4.How do we localize applications using System.Speech to take
advantage of 
> > > the eight languages supported out of the box?
> > > 
> > > 5.SAPI 5.1 seems to want to reset the microphone settings
(particularly 
> > > volume) every time the engine is started; however, these settings
are never 
> > > right (for example, our applications are used in helicopters
while in flight 
> > > -- this requires different settings that the average user).Does
System.Speech 
> > > provide controls to embed the microphone control directly into
our 
> > > applications, so that users can adjust the speech settings
without needing 
> > > direct access to the control panel? What permissions issues arise
from trying 
> > > to set the microphone volume programmatically in a
constrained-permissions 
> > > environment?
> > > a.Same question but for controls for speech
accuracy/responsiveness settings.
> > > 
> > > Thanks in advance,