Mispronunciation Detection Without L2 Pronunciation Dataset in Low-Resource Setting: A Case Study in Finland Swedish

Nhan Phan, Mikko Kuronen, Maria Kautonen, Riikka Ullakonoja, Anna von Zansen +4 more
6/1/2025
cs.CLcs.SDeess.AS

No official code found. Best implementation: aalto-speech/FinSwedish (0 stars, CC-BY-4.0, last commit 2026-04-23).

Abstract

Mispronunciation detection (MD) models are the cornerstones of many language learning applications. Unfortunately, most systems are built for English and other major languages, while low-resourced language varieties, such as Finland Swedish (FS), lack such tools. In this paper, we introduce our MD model for FS, trained on 89 hours of first language (L1) speakers' spontaneous speech and tested on 33 minutes of L2 transcribed read-aloud speech. We trained a multilingual wav2vec 2.0 model with entropy regularization, followed by temperature scaling and top-k normalization after the inference to better adapt it for MD. The main novelty of our method lies in its simplicity, requiring minimal L2 data. The process is also language-independent, making it suitable for other low-resource languages. Our proposed algorithm allows us to balance Recall (43.2%) and Precision (29.8%), compared with the baseline model's Recall (77.5%) and Precision (17.6%).

View on arXivView PDF

Code Implementations(1)

aalto-speech/FinSwedishCommunity100%
CC-BY-4.0

Repository for Finland Swedish Mispronunciation Detection Without L2 Pronunciation Dataset

00Jupyter NotebookMay 19, 20256 months agoCC-BY-4.0

Similar Papers

Mispronunciation Detection and Diagnosis Without Model Training: A Retrieval-Based Approach

87% match2025

Data-Driven Mispronunciation Pattern Discovery for Robust Speech Recognition

84% match2025

English Pronunciation Evaluation without Complex Joint Training: LoRA Fine-tuned Speech Multimodal LLM

81% match2025

Towards Efficient and Multifaceted Computer-assisted Pronunciation Training Leveraging Hierarchical Selective State Space Model and Decoupled Cross-entropy Loss

79% match1 repo152025

Non-native Children's Automatic Speech Assessment Challenge (NOCASA)

77% match2025

When Does Language Transfer Help? Sequential Fine-Tuning for Cross-Lingual Euphemism Detection

77% match2025

Cite this paper

@article{phan2025mispronunciation,
  title  = {Mispronunciation Detection Without L2 Pronunciation Dataset in Low-Resource Setting: A Case Study in Finland Swedish},
  author = {Nhan Phan and Mikko Kuronen and Maria Kautonen and Riikka Ullakonoja and Anna von Zansen and Yaroslav Getman and Ekaterina Voskoboinik and Tamás Grósz and Mikko Kurimo},
  year   = {2025},
  eprint = {2506.01156},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2506.01156v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2506.01156.svg)](https://codeofpaper.com/paper/2506.01156)

Discussion