Is there a way to conduct conditional logic in grxml?

Question

I am developing a grammar for Speech recognition and have hit a brick wall.

If a customer does not know the answer to a prompt and says "don't Know" and other customers do know the answer, I am looking at separating these out.

For Example, if an insurance number is being requested: AB112233C and the user either knows this or does not.

I want to take a particular action in the application if the user doesn't know.

I am using NUANCE as the ASR.

<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<grammar xmlns="http://www.w3.org/2001/06/grammar" 
xml:lang="en-GB" root="_alpha" version="1.0" 
mode="voice" tag-format="swi-semantics/1.0">
<rule id="_alpha" scope="public">
    <ruleref uri="#alpha"/>
    <tag>

            /* Only some letters allowed in prefix and suffix */
        var alpharegex = /^([A-CEGHJ-PR-TW-Z]){1}([A-CEGHJ-NPR-TW-Z]){1}([0-9]){2}([0-9]){2}([0-9]){2}([A-D ]){1}?$/;
        if ( !alpharegex.test(nino.out) ) SWI_disallow = 1;

        SWI_meaning = alpha.out;
    </tag>
</rule>
<rule id="alpha">
    <ruleref special="GARBAGE"/>
    <ruleref uri="#prefix"/><tag>out = prefix.out</tag>
    <ruleref uri="#digits"/><tag>out += digits.out</tag>
    <ruleref uri="#suffix"/><tag>out += suffix.out</tag>
</rule>
<rule id="prefix">
    <tag>out = '';</tag>
    <item repeat="2">
        <one-of>
            <item><tag>out += 'A'</tag>a</item>
            <item><tag>out += 'B'</tag>b</item>
            <item><tag>out += 'C'</tag>c</item>
            <item><tag>out += 'E'</tag>e</item>
            <item><tag>out += 'G'</tag>g</item>
            <item><tag>out += 'H'</tag>h</item>
            <item><tag>out += 'J'</tag>j</item>
            <item><tag>out += 'K'</tag>k</item>
            <item><tag>out += 'L'</tag>l</item>
            <item><tag>out += 'M'</tag>m</item>
            <item><tag>out += 'N'</tag>n</item>
            <item><tag>out += 'O'</tag>o</item>
            <item><tag>out += 'P'</tag>p</item>
            <item><tag>out += 'R'</tag>r</item>
            <item><tag>out += 'S'</tag>s</item>
            <item><tag>out += 'T'</tag>t</item>
            <item><tag>out += 'W'</tag>w</item>
            <item><tag>out += 'X'</tag>x</item>
            <item><tag>out += 'Y'</tag>y</item>
            <item><tag>out += 'Z'</tag>z</item>
        </one-of>
    </item>
    <tag>
        /* alpha can't start with any of these */
        var badPrefixes = /^(BG|GB|KN|NK|NT|TN|ZZ)/;
        if ( badPrefixes.test(out) ) SWI_disallow = 1;
    </tag>
</rule>
<rule id="digits">
    <tag>out = '';</tag>
    <item repeat="6">
        <one-of>
            <item><tag>out += '0'</tag>oh</item>
            <item><tag>out += '0'</tag>zero</item>
            <item><tag>out += '1'</tag>one</item>
            <item><tag>out += '2'</tag>two</item>
            <item><tag>out += '3'</tag>three</item>
            <item><tag>out += '4'</tag>four</item>
            <item><tag>out += '5'</tag>five</item>
            <item><tag>out += '6'</tag>six</item>
            <item><tag>out += '7'</tag>seven</item>
            <item><tag>out += '8'</tag>eight</item>
            <item><tag>out += '9'</tag>nine</item>
        </one-of>
    </item>
</rule>
<rule id="suffix">
    <item repeat="0-1">
        <one-of>
            <item><tag>out = 'A'</tag>a</item>
            <item><tag>out = 'B'</tag>b</item>
            <item><tag>out = 'C'</tag>c</item>
            <item><tag>out = 'D'</tag>d</item>
        </one-of>
    </item>
</rule>
</grammar>

What grammar do you have already? – Nikolay Shmyrev May 22 '18 at 16:59 — Nikolay Shmyrev, May 22 '18 at 16:59
@NikolayShmyerv I have added the code for ease. – ArcherGilly May 23 '18 at 16:31 — ArcherGilly, May 23 '18 at 16:31

score 2 · Accepted Answer · answered May 24 '18 at 13:20

Add a separate branch for "i don't know". You can use rulerefs inside one-of:

<rule id="alpha">
    <one-of>
       <item>
           <ruleref special="GARBAGE"/>
           <ruleref uri="#prefix"/><tag>out = prefix.out</tag>
           <ruleref uri="#digits"/><tag>out += digits.out</tag>
           <ruleref uri="#suffix"/><tag>out += suffix.out</tag>
       </item>
       <item>i don't know</item>
    </one-of>
</rule>

Overall, it is better to avoid complex grammars and rules. This is 2018, time to ditch Nuance. Try Google's Dialogflow or open source speech recognizer, you'll be surprised how good is it.

Is there a way to conduct conditional logic in grxml?

1 Answers1