Taiwan university creates AI tool that improves speaking skills

台灣大學研發出可提升口語表達能力的 AI 工具

TAIPEI (Taiwan News) — National Taiwan University of Science and Technology has developed an AI-powered system to help students improve their speaking skills through real-time feedback.

The tool, featuring speech recognition and language processing technologies to support self-directed learning and job training, received an award from the National Science and Technology Council. It was jointly developed by three professors from the university’s vocational education center, digital learning graduate institute, and computer science department, per CNA.

The university said the tool automatically analyzes oral presentations, offering suggestions on content, speaking rate, and tone. Users simply record and upload their speech and the system generates feedback within seconds.

The school noted the system’s built-in model for understanding spoken Mandarin can interpret speech at a secondary school level. Its integrated tone recognition model also achieved strong performance at Interspeech, the world’s largest conference on spoken language processing technology.

Chen Kuan-yu (陳冠宇), one of the professors involved in developing the system, said the team plans to expand the language model. He added that collecting a large volume of specialized terms remains a key challenge in optimizing the system.

Since April, National Taipei University of Technology has integrated an AI-powered tool into its English-speaking courses. The system offers real-time feedback, audio playback, and learning progress tracking to monitor students’ development.

After six weeks of implementation, more than half of the participants showed improvement in pronunciation and fluency, the university added.