Skip to main navigation Skip to search Skip to main content

Class-Level Error Analysis and Systematic Evaluation of CNN-Based Models for French Mispronunciation Detection

  • Universidad de Guadalajara
  • Centro Universitario de los Valles

Research output: Chapter in Book/Report/Conference proceeding › Published Conference contribution › peer-review

Abstract

Accurate pronunciation assessment remains a critical challenge in second language learning, particularly for French, where subtle phonological differences can significantly affect intelligibility. Although deep learning approaches have shown promising results in mispronunciation detection, the impact of architectural and hyperparameter variations on model behavior is not yet fully understood, especially at the class level. This study presents a systematic experimental analysis of convolutional neural network (CNN) configurations for French mispronunciation detection using a balanced dataset of Spanish-speaking learners annotated into four pronunciation categories. The analysis includes controlled variations in hyperparameters, the incorporation of Batch Normalization, increased model capacity, label smoothing, and extended training up to 200 epochs. Results show that the baseline CNN achieves strong and stable performance, while most modifications do not produce consistent improvements in macro F1-score. Batch Normalization enhances training stability but has limited impact on generalization, and increasing model capacity yields only marginal gains. In contrast, label smoothing degrades performance. Extended training experiments further reveal early convergence and clear overfitting behavior. A detailed class-level analysis indicates that the primary limitation lies in the confusion between phonetically similar error categories, particularly Error Class 1 and Error Class 2. These findings suggest that improving mispronunciation detection requires focusing on feature representation and class discrimination rather than increasing model complexity.

Original languageEnglish
Title of host publicationArtificial Intelligence – COMIA 2026 - 18th Mexican Congress, 2026, Proceedings
EditorsHiram Ponce, Hugo Jair Escalante, Roberto A. Vazquez
Pages46-58
Number of pages13
DOIs
Publication statusPublished - 2027
Event18th Mexican Congress on Artificial Intelligence, COMIA 2026 - Cuernavaca, Mexico
Duration: 25 May 2026 → 29 May 2026

Publication series

NameCommunications in Computer and Information Science
Volume3096 CCIS
ISSN (Print)1865-0929
ISSN (Electronic)1865-0937

Conference

Conference18th Mexican Congress on Artificial Intelligence, COMIA 2026
Country/TerritoryMexico
CityCuernavaca
Period25/05/26 → 29/05/26

Keywords

  • Class-level analysis
  • Convolutional neural networks
  • French mispronunciation detection
  • Hyperparameter study
  • Mel-spectrograms

Fingerprint

Dive into the research topics of 'Class-Level Error Analysis and Systematic Evaluation of CNN-Based Models for French Mispronunciation Detection'. Together they form a unique fingerprint.

Cite this