GotSpeech.NET

The online community for Microsoft Speech Server developers
Welcome to GotSpeech.NET Sign in | Join | Help
in Search
Computer-Talk

Recognition Accuracy with Weight and Wildcards

Last post 08-04-2008, 6:07 PM by JohnC. 0 replies.
Sort Posts: Previous Next
  •  08-04-2008, 6:07 PM 6592

    Recognition Accuracy with Weight and Wildcards

    How can the recognition be improved with weight and wildcards.  I have the varying case to stand out while testing.  I am using SAPI 5.1 on XP.

    I have the following grammar file code

    <GRAMMAR LANGID="409">
        <RULE NAME="what attracted you to NLM." TOPLEVEL="ACTIVE">
            <O>What</O>
            <O>yOu</O>
            <O>to</O>
            <O>NLM</O>
            <LIST>
                <PHRASE>AtTractED you to NLM</PHRASE>
                <PHRASE>... the library</PHRASE>
            </LIST>
        </RULE>

    </GRAMMAR>
     

                <PHRASE Weight=".06">...What AtTractED you to NLM</PHRASE>
                <PHRASE Weight=".04">What attracted</PHRASE>
                <PHRASE Weight=".04">attracted you</PHRASE>
                <PHRASE Weight=".04">attraction</PHRASE>
                <PHRASE Weight=".04">what attracted you to</PHRASE>
                <PHRASE Weight=".04">attracted you to NLM</PHRASE>
                <PHRASE Weight=".04">attracted you to</PHRASE>
                <PHRASE Weight=".04">to nlm</PHRASE>
     

     

    Filed under:
View as RSS news feed in XML