VoiceXML Connector
VoiceXML Connector is a Nuance microservice for running Mix dialogs in a Speech Suite deployment. It translates dialog models (created in the Nuance Mix Platform) into VoiceXML flows and serves them to the Speech Suite environment. This document describes how to use VoiceXML Connector, including installation and configuration if necessary.
When a session begins, your client application uses a <subdialog> to call VoiceXML Connector. The Connector passes the call to the Nuance Dialog service, which retrieves a dialog model. The Connector renders the dialog as VoiceXML, which executes using Speech Suite resources.
This video introduces VoiceXML Connector and shows a runtime telephone call. In the Nuance-hosted environment, the callflow is the same but the services run in Nuance Cloud IVR:
Use case scenarios
Any VoiceXML application built for Speech Suite can use the Connector to invoke a Mix dialog model. The VoiceXML application can be full-featured or an empty stub. Basic scenarios:
- Create a Mix dialog model and invoke it from an IVR stub application. New IVR sessions connect immediately to the model which drives the entirety of the conversation using Speech Suite and other resources.
- Extend an existing Speech Suite IVR application to invoke a Mix dialog model. In this scenario, your IVR application co-exists with the dialog model. For example, the IVR application might use the dialog model for a new module that was not handled previously, or you might migrate an existing application to Mix dialogs one module at a time.
Limitations
When building dialog models, only use Mix features that are supported by Speech Suite. Because the Mix platform offers functionality to a broad range of dialog applications, it includes features that are not applicable for Speech Suite platforms and cannot execute.
For details about setting up projects for IVR development see Integrating with Speech Suite in the Nuance Mix documentation.
Text-to-speech processing
In a Nuance-hosted environment, you can replace Vocalizer TTS with the URL of a Microsoft Neural TTS service. See Microsoft Neural TTS configuration.
Recommendations for Vocalizer:
- Specify the TTS language and voice in the Dialog service, and VoiceXML Connector will use the corresponding values in the Speech Suite TTS service. Speech Suite ignores TTS gender or model settings in the Dialog service (because they are not applicable).
- Optional. You can override the voice setting in the Dialog service by specifying ttsVoiceName in the VoiceXML Connector start request.
Product FAQ
Q: What is the basic concept of this product?
VoiceXML Connector expands the traditional (historical) Speech Suite platform to enable processing of cutting-edge Mix dialogs. VoiceXML Connector is the only way to run a dialog on a Speech Suite platform.
Q: Where do these cutting-edge dialogs come from?
The dialogs are created with Mix, exported to an artifact server, and made available through the Dialog service.
Q: What are the software dependencies?
VoiceXML Connector requires an IVR platform that includes Speech Suite and a VXML browser. (You can use any VXML browser that works with Speech Suite, including the Platform Add-on browser from Nuance.)
VoiceXML Connector also requires access to an application server and the Nuance Dialog service.
You also need a Kubernetes cluster or a Tomcat installation.
Q: What is the lowest overhead installation for evaluation purposes?
Probably the WAR file installation.
Q: So I need to install the Dialog service too?
Not necessarily. You have a choice. You can install the Dialog service and deploy artifacts on your own hosts or you can configure access to a Nuance-hosted Dialog service. You make this decision when completing a sales agreement with Nuance. The practical difference when you host your own installation of Dialog service (in addition to operational considerations) is that you'll need to export your dialogs from the Mix environment and deploy them to your artifact server. With a Nuance-hosted Dialog service, the dialog deployment is more simple.
Q: Are all the components microservices?
No. VoiceXML Connector and Dialog service are microservices installed on Kubernetes/Helm. You can also install VoiceXML Connector on an appserver. Speech Suite is not a microservice.
Abbreviations and terminology
Frequent abbreviations:
Term | Full meaning |
---|---|
AI | Artificial Intelligence |
ASR | Automatic Speech Recognition |
CCSS |
Customer Configuration Storage Service, a component of the Dialog service. |
Connector | Informal name in this document for VoiceXML Connector. |
client application |
The application executed on the IVR platform. |
Mix | A Nuance platform for creating conversational dialogs and related artifacts. |
Dialog service |
Alternative name for DLGaaS, Nuance's omni-channel conversation engine. |
DLM | Domain Language Model |
IVR | Interactive Voice Response |
NLE and NLU | Natural Language Engine and Natural Language Understanding |
SaaS | Software as a Service |
self-hosted | Installed in a private cloud, and sometimes called on-premise. Contrast with Nuance-hosted. |
TTS | Text to speech |
URL | Uniform Resource Locator. Client applications use URLs and URNs to load specific Mix resources. |
URN | Uniform Resource Name. Client applications use URLs and URNs to load specific Mix resources. |
VXML |
VoiceXML. The Connector renders VoiceXML 2.1 by default. You can change the version in the configuration. |