For the Language Model;

Sestek releases a general model for every language. The success rate of recognition might be low because the content of this model is rather broad. In order to increase the success rate, the content of the model should be narrowed. For example, the model to be used for an operation in a banking-related recognition should not be created with a content in finance and company’s domain.

For Grammar;

Success rates of smaller grammars are always higher. As the number of items in a grammar increases, so does the chances of misrecognition.

Related Products: Conversational IVR 1 (NLP+SR), Conversational IVR 2 (NLP+SR+Orhestrator)