In vocational education, students mainly acquire knowledge and practical skills, with limited chances to present, leading to weaker speaking abilities and possible career setbacks. This project proposes an Intelligent Oral Presentation Assistance System, offering feedback on content difficulty, content coverage, speech rate, emotion, and comprehensive feedback. It allows students to practice extensively and receive real-time feedback to improve oral presentation skills.

The Intelligent Oral Presentation Assistance System includes three self-developed models:
 
 1. Listenability Model: Designed to predict the difficulty of Chinese spoken content, which achieves an accuracy and adjacent accuracy of 83.18% and 92.30%, respectively.
 
 2. Emotion Recognition Model: Based on speech features to predict emotions, this model ranked in the top 20 globally at Interspeech 2025.
 
 3. Large Language Model: Fine-tuned to enable the system to provide real-time, personalized feedback.

The "Intelligent Oral Presentation Assistance System" not only enhances students’ speaking skills but also serves as a tool for quantifying candidates’ oral abilities in job interviews. The system’s various models can be applied independently. For instance, the listenability model supports personalized learning, the emotion recognition model facilitates counseling and customer service, and the content coverage model assists with summary writing and assessment.