Utilizing CNN architectures for non-invasive diagnosis of speech disorders – further experiments and insights

This research investigated the application of deep neural networks for diagnosing diseases that affect the voice and speech mechanisms through the non-invasive analysis of vowel sound recordings. Using the Saarbruecken Voice Database, the voice recordings were converted to spectrograms to train the...

Full description

Saved in:
Bibliographic Details
Main Authors: Filip Ratajczak, Mikołaj Najda, Kamil Szyc
Format: Article
Language:English
Published: Polish Academy of Sciences 2025-07-01
Series:International Journal of Electronics and Telecommunications
Subjects:
Online Access:https://journals.pan.pl/Content/135742/14_5010_Ratajczak_L_sk.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849470193412079616
author Filip Ratajczak
Mikołaj Najda
Kamil Szyc
author_facet Filip Ratajczak
Mikołaj Najda
Kamil Szyc
author_sort Filip Ratajczak
collection DOAJ
description This research investigated the application of deep neural networks for diagnosing diseases that affect the voice and speech mechanisms through the non-invasive analysis of vowel sound recordings. Using the Saarbruecken Voice Database, the voice recordings were converted to spectrograms to train the models, specifically focusing on the vowels /a/, /u/, and /i/. The study used Explainable Artificial Intelligence (XAI) methodologies to identify essential features within these spectrograms for pathology identification, with the aim of providing medical professionals with enhanced insight into how diseases manifest in sound production. The F1 Score performance evaluation showed that the DenseNet model scored 0.70 ± 0.03 with a top of 0.74. The findings indicated that neither vowel selection nor data augmentation strategies significantly improved model performance. Additionally, the research highlighted that signal splitting was ineffective in enhancing the models’ ability to extract features. This study builds on our previous research [1], offering a more comprehensive understanding of the topic.
format Article
id doaj-art-ddaaf5825cd64186b0ee97bd9b3cdeea
institution Kabale University
issn 2081-8491
2300-1933
language English
publishDate 2025-07-01
publisher Polish Academy of Sciences
record_format Article
series International Journal of Electronics and Telecommunications
spelling doaj-art-ddaaf5825cd64186b0ee97bd9b3cdeea2025-08-20T03:25:12ZengPolish Academy of SciencesInternational Journal of Electronics and Telecommunications2081-84912300-19332025-07-01vol. 71No 3https://doi.org/10.24425/ijet.2025.153621Utilizing CNN architectures for non-invasive diagnosis of speech disorders – further experiments and insightsFilip Ratajczak0Mikołaj Najda1Kamil Szyc2Faculty of Information and Communication Technology, Wrocław University of Science and Technology, Wrocław, PolandInstitute of Data Science, Maastricht University, The NetherlandsFaculty of Information and Communication Technology, Wrocław University of Science and Technology, Wrocław, PolandThis research investigated the application of deep neural networks for diagnosing diseases that affect the voice and speech mechanisms through the non-invasive analysis of vowel sound recordings. Using the Saarbruecken Voice Database, the voice recordings were converted to spectrograms to train the models, specifically focusing on the vowels /a/, /u/, and /i/. The study used Explainable Artificial Intelligence (XAI) methodologies to identify essential features within these spectrograms for pathology identification, with the aim of providing medical professionals with enhanced insight into how diseases manifest in sound production. The F1 Score performance evaluation showed that the DenseNet model scored 0.70 ± 0.03 with a top of 0.74. The findings indicated that neither vowel selection nor data augmentation strategies significantly improved model performance. Additionally, the research highlighted that signal splitting was ineffective in enhancing the models’ ability to extract features. This study builds on our previous research [1], offering a more comprehensive understanding of the topic.https://journals.pan.pl/Content/135742/14_5010_Ratajczak_L_sk.pdfvoice disorder diagnosisvowel sound analysisconvolutional neural networks (cnns)explainable artificial intelligence (xai)
spellingShingle Filip Ratajczak
Mikołaj Najda
Kamil Szyc
Utilizing CNN architectures for non-invasive diagnosis of speech disorders – further experiments and insights
International Journal of Electronics and Telecommunications
voice disorder diagnosis
vowel sound analysis
convolutional neural networks (cnns)
explainable artificial intelligence (xai)
title Utilizing CNN architectures for non-invasive diagnosis of speech disorders – further experiments and insights
title_full Utilizing CNN architectures for non-invasive diagnosis of speech disorders – further experiments and insights
title_fullStr Utilizing CNN architectures for non-invasive diagnosis of speech disorders – further experiments and insights
title_full_unstemmed Utilizing CNN architectures for non-invasive diagnosis of speech disorders – further experiments and insights
title_short Utilizing CNN architectures for non-invasive diagnosis of speech disorders – further experiments and insights
title_sort utilizing cnn architectures for non invasive diagnosis of speech disorders further experiments and insights
topic voice disorder diagnosis
vowel sound analysis
convolutional neural networks (cnns)
explainable artificial intelligence (xai)
url https://journals.pan.pl/Content/135742/14_5010_Ratajczak_L_sk.pdf
work_keys_str_mv AT filipratajczak utilizingcnnarchitecturesfornoninvasivediagnosisofspeechdisordersfurtherexperimentsandinsights
AT mikołajnajda utilizingcnnarchitecturesfornoninvasivediagnosisofspeechdisordersfurtherexperimentsandinsights
AT kamilszyc utilizingcnnarchitecturesfornoninvasivediagnosisofspeechdisordersfurtherexperimentsandinsights