Bipolar disorder is a common psychiatric disorder. At present， the diagnosis of bipolar disorder mainly relies on clinical experience， and its early recognition rate and diagnostic rate are relatively low. There are differences in language signal characteristics between individuals with bipolar disorder and healthy individuals， which change with patients' emotions and disease state， and have the potential to become objective evaluation indicators for bipolar disorder. This paper introduces the basic concepts and methods of speech recognition technology， as well as the research progress of speech recognition in bipolar disorder， with the aim of identifying speech feature categories that reflect bipolar disorder and providing a basis for establishing an objective evaluation system for bipolar disorder.