Information Technology Reference
In-Depth Information
Speech Recognition
To facilitate speech recognition, UCCX (CRS) comes with a Media Resource Control Protocol (MRCP, RFC 4463) Automated
Speech Recognition (ASR) component. This component, along with a separate ASR server provided by a third-party vendor, such as
Nuance or Scansoft, can be used to enable speech recognition.
Figure 6-2
illustrates connectivity between an UCCX cluster and an ASR server.
FIguRE 6-2
Connectivity Between
an UCCX Cluster and
an ASR Server
Cisco
CUCM
MRCP/RTSP
Media Streams (RTP)
UCCX Cluster
(MRCP Client)
ASR Server
(MRCP Server)
PSTN
V
Agent's
Phones/PCs
Supervisor's
Phones/PCs
MRCP is a mechanism that allows a client device that requires audio stream processing to control processing resources, such as ASR
and TTS servers, and instructs them to perform tasks such as speech recognition and text-to-speech conversion. MRCP relies on
the Real Time Streaming Protocol (RTSP) or SIP as a control protocol, which is responsible for setting up and controlling sessions.
RTSP/SIP is also responsible for setting up media streams between the client and server using RTP.
UCCX (CRS) supports ASR via the MRCP ASR subsystem.
