Hi, assume I have a knowledge base attached to a query-tool. Question: Do the voice replacements (if any) are applied to the result from query-tool? Or query-tool response are "spoken" exactly as in the knowledge base content ?