Nonlinear Network Speech Recognition Structure in a Deep Learning Algorithm.

Area of Science:

• Artificial Intelligence
• Computational Linguistics
• Educational Technology

Background:

• Globalization is increasing English language learning in China.
• A significant shortage of qualified English teachers hinders effective instruction.
• Existing methods for pronunciation feedback are often limited.

Purpose of the Study:

• To propose and evaluate a deep learning-based algorithm for English pronunciation assessment.
• To enhance learners' ability to distinguish and correct phonemic errors.
• To provide automated, impartial feedback for oral English improvement.

Main Methods:

• A deep learning algorithm was developed, integrating nonlinear network identification and neural network models.
• The algorithm was used to evaluate English pronunciation quality and speech recognition.
• Experimental results were analyzed using metrics like concordance and Pearson correlation.

Main Results:

• Machine and manual intonation evaluation showed 80% concordance.
• Adjacent intonation evaluation achieved a 98.33% concordance rate.
• A Pearson correlation coefficient of 0.627 indicated the technique's reliability.

Conclusions:

• The proposed deep learning model for English pronunciation and speech identification is effective and dependable.
• The algorithm provides timely, accurate, and impartial guidance for learners.
• This technology can significantly enhance oral English learning capacity by addressing phonemic errors.