Developing Speech Grammars

Nuance Recognizer Version 9 and OpenSpeech Recognizer offer native support of the XML Grammar Format of the Speech Recognition Grammar Specification. On the surface, this is a simple way to write speech recognition grammars, but the simple format hides complex decisions that affect recognition.

This course introduces best practices for creating an XML grammar as it describes the role grammars play in the speech recognition process. It discusses the rules and rule references necessary to build XML Grammars and shows participants how to test basic XML grammars using the Nuance parseTool and test_parser tools. Participants learn how to streamline semantic interpretation and pass concise semantic values to applications. The class also discusses strategies to optimize the performance of the Nuance Recognizer Version 9 or OpenSpeech Recognizer.

In addition to learning about important grammar topics, students learn to leverage built-in grammars to handle common collection contexts and to help improve accuracy by creating lexicons of additional or alternate pronunciations to match caller input.

• Build basic XML grammars in accord with accepted standards
• Demonstrate how rules and rule references are used to build XML grammars
• Create functional grammars using all of the basic elements
• Apply semantic interpretation principles within your grammars to return the appropriate meaning to the application
• Use Nuance Recognizer (NR9) tools to test XML grammars
• Create user dictionaries and use in conjunction with system dictionaries

• Grammar developers
• Application Developers
• Speech Scientists
• User Interface Designers
• Project Managers

1. Introduction to grammars and XML
2. Basic GRXML grammar syntax - SRGS
3. Rules and rule references
4. Semantic Interpretation and special keys
5. Pronunciations and dicationaries

• Introduction to Speech Recognition (eCampus)
• Fundamentals of Voice User Interface Design (eCampus)
• Basic computer programming background