Article: Fast capture of English pronunciation errors using deep learning approaches Journal: International Journal of Business Intelligence and Data Mining (IJBIDM) 2026 Vol.28 No.10 pp.165 - 182 Abstract: This paper proposes a fast pronunciation error detection method based on deep learning to address the problems of traditional methods relying on manual features and insufficient sequence modelling ability in English oral pronunciation error detection, which limits detection accuracy and efficiency. Firstly, pre-process the speech signal and construct a Mel spectrogram to preserve the complete spectral structure; Secondly, utilising convolutional neural networks to automatically extract deep discriminative features to avoid information loss from manual features; Finally, an end-to-end sequence modelling framework combining bidirectional long short-term memory networks and connected temporal classification is constructed to achieve accurate detection and localisation of phoneme level errors. The experimental results show that this method has an accuracy rate of over 95% in detecting pronunciation errors, a recall rate of over 0.9, and significantly better detection time than the comparative methods, verifying its comprehensive advantages in detection accuracy and computational efficiency. Inderscience Publishers - linking academia, business and industry through research

Title: Fast capture of English pronunciation errors using deep learning approaches

Authors: Meizhen Ji

Addresses: School of International Education, Jiangsu Vocational College of Electronics and Information, Huai'an, 223003, China

Abstract: This paper proposes a fast pronunciation error detection method based on deep learning to address the problems of traditional methods relying on manual features and insufficient sequence modelling ability in English oral pronunciation error detection, which limits detection accuracy and efficiency. Firstly, pre-process the speech signal and construct a Mel spectrogram to preserve the complete spectral structure; Secondly, utilising convolutional neural networks to automatically extract deep discriminative features to avoid information loss from manual features; Finally, an end-to-end sequence modelling framework combining bidirectional long short-term memory networks and connected temporal classification is constructed to achieve accurate detection and localisation of phoneme level errors. The experimental results show that this method has an accuracy rate of over 95% in detecting pronunciation errors, a recall rate of over 0.9, and significantly better detection time than the comparative methods, verifying its comprehensive advantages in detection accuracy and computational efficiency.

Keywords: deep learning; spoken English; pronunciation error; error capture; convolutional neural network; CNN; connection timing classification.

DOI: 10.1504/IJBIDM.2026.156189

International Journal of Business Intelligence and Data Mining, 2026 Vol.28 No.10, pp.165 - 182

Received: 25 Feb 2026
Accepted: 02 Jul 2026
Published online: 07 Sep 2026 *

Title: Fast capture of English pronunciation errors using deep learning approaches

Keep up-to-date