I've read the other posts re nBest and Alternates, but we're still having problems. I've also checked out MSDN recognitionresult.alternates, but we only get one alternate no matter how many homophones we have in the grammar.
I've also set max results -- this.speechrecognizer.maxresults=5;
Our application is a pretty straightforward speech attendant, whereby the caller says the name of the person they want to reach. We're developing in VS2005/WWF and have some experience with speech server --- so this one is KILLING me. So to boil the issue down, we're using very simple words in our grammar now. Three of the entries are "to", "too", and "two". When we say the word "to" or "two" or "too" (all sound the same!) the recognition result is "two" and the first alternate is "two", as is designed.
However, there are NO further alternates listed, even though they're in the grammar, as illustrated below:
</item>
<item>
<?MS_Grammar_Editor GroupWrap?>
<one-of>
<item>to</item>
</one-of>
<tag>$._value = "to"
$.extension={}; $.extension._value="9999";</tag>
</item>
<item>
<?MS_Grammar_Editor GroupWrap?>
<one-of>
<item>two</item>
</one-of>
<tag>$._value = "two"
$.extension={}; $.extension._value="6464";</tag>
</item>
<item>
<?MS_Grammar_Editor GroupWrap?>
<one-of>
<item>too</item>
</one-of>
<tag>$._value = "too"
$.extension={}; $.extension._value="8888";</tag>
I'm calling into the system and debugging via a live call. I'm not using the onboard sip phone debugger. I have a debugger screenshot if anyone wants to see it. I'm very grateful for any suggestions.
John Michne
DiRAD Technologies
john.michne@dirad.com