Overview
|
 Inference Tuner is an award winning development tool designed to help developers automatically generate grammar files using learning-by-example artificial intelligence techniques. Tuner can be used independently of the method used to create the speech application's VoiceXML dialogs, so that developers can still use their existing tools to create the VoiceXML for a solution and then develop the grammar files using Tuner.
Using Tuner, grammars (both simple and sophisticated natural language grammar files) can be generated in a fraction of the time taken by traditional techniques; and can be deployed on a wide variety of speech recognition platforms. This tool is ideal for developer groups attempting to manage and maintain existing speech applications which may have sub-optimal grammars or where the grammar files are difficult and time consuming to maintain.
|
How it Works?
|
Inference Tuner leverages the core of the Grammatical Inference engine, and is designed specifically for speech developers who prefer to code VoiceXML with a 3rd party tool but would like to make use of the grammar training capabilities of Talkback. It is used as a stand alone grammar generation capability with the ability to integrate with other speech recognition IDE products.
|
| Using Inference Tuner, developers are able to easily and intuitively create sophisticated speech grammars by following these simple steps: |
 |
|
Inference Tuner V2.0 Features
|
Grammar Specification Editor
|
| | - XML Editor with templates
- Validator
|
| - Builds starting grammars
- Combinator sub-tool: mixed initiative
|
| | | - Interactive vs batch mode
- Importing transcripts
|
| - Grammars generated in multiple formats
- SLM grammar generation (Nuance format)
|
|
|
|
For more information download the 'Inference Tuner Information Sheet'.
|
