Properties in MRCP/VXML/API
One challenge when developing applications using standards such as MRCP and VoiceXML is that the names of similar parameters often differ between the standards. This is particularly important when tuning voice activity detection (VAD) parameters.
The following table presents a comparison of VAD properties in MRCP and their equivalent names in VoiceXML and the Speech Engine API.
In some cases the parameters do not have direct matches, so the exact mappings may differ depending on how a platform interprets one specification relative to the other. Consider this table an informational guide rather than an official interpretation of the standards.
MRCP | VoiceXML | API |
Sensitivity-Level 1 | sensitivity 1 | STREAM_PARM_VAD_VOLUME_SENSITIVITY |
Speech-Complete-Timeout | completetimeout | STREAM_PARM_VAD_EOS_DELAY |
Speech-Incomplete-Timeout 2 | incompletetimeout 2 | Partially supported 2 |
com.lumenvox.end-of-speech-timeout | N/A | STREAM_PARM_END_OF_SPEECH_TIMEOUT |
No-Input-Timeout | timeout | STREAM_PARM_BARGE_IN_TIMEOUT |
Confidence-Threshold | confidencelevel | N/A (handled by application) |
Speed-Vs-Accuracy | speedvsaccuracy | N/A |
N-Best-List-Length | maxnbest | PROP_EX_MAX_NBEST_RETURNED |
Recognition-Timeout | maxspeechtimeout | N/A |
Waveform-URL | N/A | N/A |
DTMF-Interdigit-Timeout | interdigittimeout | N/A |
DTMF-Term-Timeout | termtimeout | N/A |
DTMF-Term-Char | termchar | N/A |
1 Note that Sensitivity-Level (sensitivity in VoiceXML) works differently in MRCP/VoiceXML than in the API. In the API, setting this value higher makes it less sensitive to barge-in. MRCP and VoiceXML are the opposite.
2 Users can elect to use the speech-incomplete-timeout value by enabling the 'use_speech_incomplete' option in the media_server.conf settings file. If enabled, the greater of speech-complete-timeout and speech-incomplete-timeout will be used.