- Contents
Automated Speech Recognition Technical Reference
ASR Server Properties
The following table lists the server properties that are common to all ASR servers. For information about configuring the properties, see Configure ASR Server Properties for MRCP.
|
Property Name |
Description |
|---|---|
|
Nuance |
|
|
|
Indicates
whether to create slots that start with |
|
|
Allows the system to send a custom session.xml configuration to the MRCP server during the SIP INVITE. For more information, see the Nuance documentation. |
|
Common Properties |
|
|
|
List of confidence mapping pairs used to map the native engine confidences into the confidence values returned in the recognition result. Used to normalize differing confidence values that the ASR engines return. Each mapping pair consists of two floating point values between 0.0 and 1.0 separated by an equal sign. The first value designates the native value that the engine returns and the second value the resulting mapped confidence for that particular engine value. The property might contain an arbitrary number of mapping points. Uses linear interpolation to map confidence values that lie between two points. When mapping confidence values, the system always assumes two implicit mapping points, 0.0=0.0 and 1.0=1.0. This method results in an identity mapping when no property is defined, and the results ensure correct handling at the boundaries. |
|
|
Boolean to allow logging of WAV data for debugging purposes. Default = False |
|
|
Minimum confidence an ASR recognition hypothesis must have for inclusion in the recognition result. Default = 0.1 |
|
|
Timeout, in seconds, to use when waiting on grammars to compile. Default = 300 (5 minutes) |
|
|
Boolean
to control the return of the utterance to the application. If
set to True, the recognition results that the
MRCP server returns contain a custom field called |
|
|
Media
format of the utterance recording. For example, using this property
with a value of audio/x-alaw-basic requests the utterance recording
in the |
|
|
Prevents the system from logging sensitive data in plain text in the IP log. Set to True for an input operation. The system sets a flag in the recognition result that suppresses tracing of the utterance and slot values in the trace logs. |
|
|
IP address or DNS name under which the CIC server sees the RTP endpoint of the ASR Server. |
|
|
Local IP address of the NIC, which the server uses to receive audio. This address must refer to the same physical device as the recocfg:AudioAdvertisedAddress property. Specify the local NIC address as a dot address. |
|
MRCP |
|
|
|
Indicates whether to include the raw NLSML result as a custom element. If this value is set to True, the system adds the <result> element of the NLSML result data to the end of the recognition result, wrapped in a <RawResult> element. Default = False. |
|
|
Sends a custom header field to the MRCP server. For example, in the Select or enter name of property text box, type recomrcp:Vendor-Specific-Parameters to send vendor-specific parameters to the MRCP server. |
|
|
Indicates whether to dump received RTP packets to a capture file for diagnostic purposes. The capture files reside in the trace directory. |
|
|
Indicates
whether to send MRCP |
|
|
Default language. Since grammar files specify the language to use, the system doesn't require this property. However, you might need a default language to register top-level built-in grammars. Note: |

