Speech recognition applications use confidence scoring to evaluate the quality of the caller input. The confidence score is an indicator of how closely the caller's utterance matches the phrases specified in the grammar.

• No Input from caller: Used when the application doesn't receive any discernible response from the caller, No Input results in a general re-prompt message, "I'm not hearing you."
• Low Confidence Recognition of caller recognition: Indicates that the application received input, but can't be sure what the caller really said. A typical message in this case is "Say again please."
• Medium Confidence Recognition: Indicates that the application received input, but needs to confirm the caller's response with a message such as "<speaking back response> Is this correct?"
• High Confidence Recognition: Used when the application receives non-ambiguous input which was scored above the high confidence score threshold. In this case, the application accepts the response and continues.