In the course of daily human interaction, speech is the most direct, convenient, and commonly used communication approach. Besides the content, speech often contains rich emotional characteristics. So speech emotion recognition is a key research direction of emotional computing. This project proposes a speech emotion recognition model based on random forest and improves its performance effectively. Considering its importance role in medicine addressed by many scholars, the project also conducts the heartbeat anomaly detection to assist doctors’ diagnosis. With the system, people can judge the emotion states through speech.