Speech-Related Components
These components provide the interface to third-party speech servers that deliver speech recognition and text-to-speech services for VoiceXML. GVP uses the Media Resource Control Protocol to communicate with these services. The MCP includes client interfaces for MRCPv1 and MRCPv2. These clients are capable of load-balancing across pools of speech servers, and service monitoring of speech servers for availability purposes. GVP also includes an MRCPv1 proxy that provides global load balancing, peak reporting, and other services.
- MRCP Proxy is an integral component that interfaces with the MRCP, Management Framework, and Operational Reporting API. The MRCP Proxy can be placed between the Media Control Platforms and the MRCPv1 resources within a GVP deployment. Deploying the MRCP Proxy enables ASR/TTS usage reporting data to be sent to the Reporting Server.
- For more information about the MRCP Proxy application, see MRCP Proxy in the GVP Deployment Guide. For various MRCP Proxy configuration options, see MRCP Proxy Options in this document.
GVP provides transparent access to MRCP services from VoiceXML. GVP will acquire and releases ASR and TTS sessions as required for proper execution of a VoiceXML applications. Text to Speech services render audio streams from text, allowing playback of dynamic text for which pre-recorded prompts may be difficult or impossible to provide. Text rendering can be controlled using the Speech Synthesis Markup Language (SSML), and can support multiple languages.
Speech Recognition services can process caller audio and, and using either pre-defined context free grammars or natural language models, return an interpretation of what the caller said. Speech recognition services support the Speech Recognition Grammar Specification (SRGS) for grammar specification, and the Natural Language Semantic Markup Language (NLSML) specification for describing the results of a recognition.