Jump to: navigation, search

GVP MCP:asr:load once per call
 
(8 intermediate revisions by 3 users not shown)
Line 4: Line 4:
 
|valid=Choose between: 0 or 1
 
|valid=Choose between: 0 or 1
 
|takeseffect=Immediately/session
 
|takeseffect=Immediately/session
−
|shortdesc=When this parameter is set to Enable, there will be only one VRM session for the entire call which could have multiple recognition sessions. If the parameter value is set to Disable, a VRM session is opened for each recognition request. The VRM session is closed when the recognition request is completed successfully or unsuccessfully (such as no match). As a result, there could be multiple VRM sessions in a call. Having multiple VRM sessions in a call could make the ASR server license usage more efficient. However, this configuration could have the following consequences:
+
|shortdesc=
 +
When this parameter is set to Enable, there will be only one VRM session for the entire call which could have multiple recognition sessions. If the parameter value is set to Disable, a VRM session is opened for each recognition request. The VRM session is closed when the recognition request is completed successfully or unsuccessfully (such as no match). As a result, there could be multiple VRM sessions in a call. Having multiple VRM sessions in a call could make the ASR server license usage more efficient. However, this configuration could have the following consequences:
 
<br />1. There will be longer delays on speech barge in.
 
<br />1. There will be longer delays on speech barge in.
 
<br />2. The save utterance data could be deleted by some recognizer servers after each VRM session. In that case, the VoiceXML application cannot refer to the saved utterance file after the recognition session.
 
<br />2. The save utterance data could be deleted by some recognizer servers after each VRM session. In that case, the VoiceXML application cannot refer to the saved utterance file after the recognition session.
 +
<br />Note: for ASR engines such as Google Speech-to-Text that need to open a VRM session for each recognition request, this option will be diasbled automatically.
 +
 
}}
 
}}
 
{{DraftOption}}
 
{{DraftOption}}
−
 
−
 
−
 
−
 
  
 
{{PubExtended}}
 
{{PubExtended}}
−
 
−
 
−
 
  
 
{{DraftExtended}}
 
{{DraftExtended}}
−
 
−
 
−
 
−
 
  
 
{{OptionCreate
 
{{OptionCreate

Latest revision as of 16:01, June 29, 2020



View in ref guide.

Edit this option          Publish this option          Clear draft content


Published Option

load_once_per_call

Default Value: 1
Valid Values: Choose between: 0 or 1
Changes Take Effect: Immediately/session


When this parameter is set to Enable, there will be only one VRM session for the entire call which could have multiple recognition sessions. If the parameter value is set to Disable, a VRM session is opened for each recognition request. The VRM session is closed when the recognition request is completed successfully or unsuccessfully (such as no match). As a result, there could be multiple VRM sessions in a call. Having multiple VRM sessions in a call could make the ASR server license usage more efficient. However, this configuration could have the following consequences:
1. There will be longer delays on speech barge in.
2. The save utterance data could be deleted by some recognizer servers after each VRM session. In that case, the VoiceXML application cannot refer to the saved utterance file after the recognition session.
Note: for ASR engines such as Google Speech-to-Text that need to open a VRM session for each recognition request, this option will be diasbled automatically.

Draft Option

No draft option

This page was last edited on June 29, 2020, at 16:01.
Comments or questions about this documentation? Contact us for support!