where ipaddress needs to be replaced with the corresponding IP address of UniMRCP server.
Initiating Recognition
In order to initiate a speech recognition, a built-in grammar "speech/transcribe" must be to be referenced from a VoiceXML application. For example:
<field>
<grammar src="builtin:speech/transcribe"/>
</field>
Processing Results
Upon a successful completion of the request, a received
QueryResult structure is transformed to XML and incorporated into the <instance> element of NLSML results sent back to the VoiceXML application. All the QueryResult fields can be referenced in the VoiceXML application by means of lastresult$.interpretation variable. For example:
<field>
<grammar src="builtin:speech/transcribe"/>
<filled>
<prompt>You said <break time="300ms"/> <value expr="lastresult$.interpretation.query_text"/> <break time="300ms"/></prompt>
<prompt><value expr="lastresult$.interpretation.fulfillment_text"/></prompt>
</filled>
</field>
Sample Application
You may check out the following sample VoiceXML application.
In order to use the application, simply point your VoiceXML browser to the URI above.
This should be enough to get started. Your questions, thoughts and ideas are welcome.