Scholarlink

This paper presents a framework aimed at improving the pronunciation accuracy of non-native French speakers, particularly those preparing for the DELF A2 exam. It leverages advanced speech recognition and phonetic analysis to identify frequent pronunciation errors at both word and phoneme levels. A graph-based knowledge base captures common mispronunciation patterns, which guide personalized pronunciation lessons focusing on prevalent errors. The system has been introduced to ten second-year French program students from Chiang Mai University, and early results indicate improvements in pronunciation accuracy, especially in vowel distinction and reducing errors with similar-sounding words. Ongoing data collection will further refine the system to address the unique pronunciation challenges faced by Thai learners. This adaptive, data-driven approach has demonstrated potential for enhancing pronunciation fluency, with future work focusing on broader scalability and application to diverse learner populations.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Personalized Pronunciation Practice Recommendation in Non-native French Speakers Using Graph-Based Analysis

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Kittipitch Kuptavanich,
  • Ratsameetip Wita,
  • Montiya Phoungsub

摘要

This paper presents a framework aimed at improving the pronunciation accuracy of non-native French speakers, particularly those preparing for the DELF A2 exam. It leverages advanced speech recognition and phonetic analysis to identify frequent pronunciation errors at both word and phoneme levels. A graph-based knowledge base captures common mispronunciation patterns, which guide personalized pronunciation lessons focusing on prevalent errors. The system has been introduced to ten second-year French program students from Chiang Mai University, and early results indicate improvements in pronunciation accuracy, especially in vowel distinction and reducing errors with similar-sounding words. Ongoing data collection will further refine the system to address the unique pronunciation challenges faced by Thai learners. This adaptive, data-driven approach has demonstrated potential for enhancing pronunciation fluency, with future work focusing on broader scalability and application to diverse learner populations.