Lei, Yining (2024) Chinese-speaking English learners' Vowel Pronunciation Error Detection. Master thesis, Voice Technology (VT).
|
PDF
MA-5558506-Y-Lei.pdf Download (1MB) | Preview |
Abstract
English is the main international language in the world today, and its scope of use is very wide. The study of English has always been valued by all countries. In my country, although many college students have good English grades, their oral ability is relatively weak. With the rapid development of Internet information technology, the use of multimedia to promote the learning of English oral pronunciation has become a hot topic in the field of assisting English oral learning. As an important element of clear and authentic pronunciation, vowels are mostly learned through classroom demonstration by teachers and imitation by students, lacking personalized and targeted pronunciation guidance. There is a lack of timely feedback, and the teaching resources and teaching methods are single. Therefore, this paper designs and implements a hybrid neural network model combining CNN and LSTM, and further improves the model performance by introducing a complex network architecture to obtain an English vowel phonetic pronunciation error recognition model based on speech recognition technology. The Mel Frequency Cepstrum Coefficient (MFCC) is used to extract the characteristics of the speech signal using speech signal processing technology. By matching it with the vowel data of the native speaker, the gap between the test speech and the standard speech is obtained as the scoring result. At the same time, the formant is extracted and compared with the vowel data of the native speaker to give specific vowel pronunciation tongue position suggestions. The English vowel phonetic mispronunciation recognition model constructed in this paper can quickly analyze a large amount of speech data and accurately identify the specific location and type of pronunciation errors. At the same time, the model can also provide personalized feedback and suggestions based on the learner's pronunciation characteristics to help learners improve their pronunciation in a targeted manner. This personalized teaching method, compared with traditional teaching methods, can better meet the needs of different learners and improve teaching effectiveness. The relevant audio and model has been uploaded to the Github (https://github.com/LeiYining/Vowel-Pronunciation-Error-Detection).
| Item Type: | Thesis (Master) |
|---|---|
| Name supervisor: | Coler, M.L. |
| Date Deposited: | 11 Jul 2024 08:33 |
| Last Modified: | 11 Jul 2024 08:33 |
| URI: | https://campus-fryslan.studenttheses.ub.rug.nl/id/eprint/510 |
Actions (login required)
 |
View Item |