SpeechRecognizer class (as opposed to DesktopRecognizer)

[David Pfeffer]

This message is kind of angled towards Robert Brown, but everyone
should feel free to help me out. =)

I have an application working with the DesktopRecognizer class.
However, I want to switch to using an in-process recognizer (and thus
the SpeechRecognizer class) so that I can have multiple recognizers on
different sound cards. However, I am not sure what to pass into the
SpeechRecognizer (RecognizerBase?)'s SetInput method. They seem to be
able to accept filenames/streams, but I'm not sure how to handle it if
I want to specify an actual sound device.

Also, will an in-process speech recognizer be able to interact with a
user if it is put in a Windows service? Ideally, recognition for my
software would begin on system boot, as soon as my service starts.

Thanks in advance,
David Pfeffer

[Robert Brown [MSFT]]

Hi David,

The short answer is that you can't attach it to a particular sound card,
only to the sound card that's marked as default in the control panel.

One of the things we're wondering about is whether you're trying to get
audio from multiple sound cards at the same time? We don't have a
particular API for this. The generic solution when you don't want to
recognize from the default device, or from a file, is to recognize from a
stream.

Let me know if this helps.

Cheers,

/Rob

--
Robert Brown
Program Manager (Speech & Natural Language)
Microsoft
http://blogs.msdn.com/robertbrown
This posting is provided "AS IS" with no warranties, and confers no rights