Validating automated speech timing methods in clinical and healthy speakers across sentence, paragraph, and monologue tasks.