Evaluating State of the Art Voice Conversion Models for Dysphonic and Electro-Larynx Speech

Benedikt Mayrhofer; Martin Hagmueller; Franz Pernkopf; Philipp Aichinger

Evaluating State of the Art Voice Conversion Models for Dysphonic and Electro-Larynx Speech

Institute of Signal Processing and Speech Communication (4420)

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

Abstract

Pathological speech, caused by dysphonia or produced via electro-larynx devices, often suffers from poor intelligibility and unnatural prosody. In this paper, we investigate the potential of four state-of-the-art voice conversion models: FreeVC, QuickVC, LLVC, and XVC for restoring healthy-sounding speech. All models are fine-
tuned on Austrian-German datasets and evaluated using objective and subjective metrics. Results show substantial gains in intelligibility, naturalness, and perceived vocal health. QuickVC, FreeVC, and XVC perform similarly and achieve the highest preference scores, exceeding unprocessed pathological speech by up to 200%. These
findings highlight the potential to improve communication for individuals with voice disorders and motivate further development of efficient, high-quality conversion systems.

Original language	English
Title of host publication	Models and Analysis of Vocal Emissions for Biomedical Applications
Subtitle of host publication	14th International Workshop
Publisher	Firenze University Press
Pages	33 - 36
ISBN (Electronic)	979-12-215-0821-5
ISBN (Print)	979-12-215-0820-8
Publication status	Published - 2025
Event	14th International Workshop on Models and Analysis of Vocal Emissions for Biomedical Applications, MAVEBA 2025 - Firenze, Italy Duration: 16 Dec 2025 → 17 Dec 2025

Conference

Conference	14th International Workshop on Models and Analysis of Vocal Emissions for Biomedical Applications, MAVEBA 2025
Country/Territory	Italy
City	Firenze
Period	16/12/25 → 17/12/25

Fields of Expertise

Information, Communication & Computing

Cite this

Evaluating State of the Art Voice Conversion Models for Dysphonic and Electro-Larynx Speech. / Mayrhofer, Benedikt ; Hagmueller, Martin ; Pernkopf, Franz et al.
Models and Analysis of Vocal Emissions for Biomedical Applications: 14th International Workshop. Firenze University Press, 2025. p. 33 - 36.

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

Mayrhofer, B , Hagmueller, M , Pernkopf, F & Aichinger, P 2025, Evaluating State of the Art Voice Conversion Models for Dysphonic and Electro-Larynx Speech. in Models and Analysis of Vocal Emissions for Biomedical Applications: 14th International Workshop. Firenze University Press, pp. 33 - 36, 14th International Workshop on Models and Analysis of Vocal Emissions for Biomedical Applications, MAVEBA 2025, Firenze, Italy, 16/12/25.

@inbook{c590e8a2071c4ed89052b6cdd4bdebd4,

title = "Evaluating State of the Art Voice Conversion Models for Dysphonic and Electro-Larynx Speech",

abstract = "Pathological speech, caused by dysphonia or produced via electro-larynx devices, often suffers from poor intelligibility and unnatural prosody. In this paper, we investigate the potential of four state-of-the-art voice conversion models: FreeVC, QuickVC, LLVC, and XVC for restoring healthy-sounding speech. All models are fine- tuned on Austrian-German datasets and evaluated using objective and subjective metrics. Results show substantial gains in intelligibility, naturalness, and perceived vocal health. QuickVC, FreeVC, and XVC perform similarly and achieve the highest preference scores, exceeding unprocessed pathological speech by up to 200\%. These findings highlight the potential to improve communication for individuals with voice disorders and motivate further development of efficient, high-quality conversion systems.",

author = "Benedikt Mayrhofer and Martin Hagmueller and Franz Pernkopf and Philipp Aichinger",

year = "2025",

language = "English",

isbn = "979-12-215-0820-8",

pages = "33 -- 36",

booktitle = "Models and Analysis of Vocal Emissions for Biomedical Applications",

publisher = "Firenze University Press",

note = "14th International Workshop on Models and Analysis of Vocal Emissions for Biomedical Applications, MAVEBA 2025 ; Conference date: 16-12-2025 Through 17-12-2025",

}

TY - CHAP

T1 - Evaluating State of the Art Voice Conversion Models for Dysphonic and Electro-Larynx Speech

AU - Mayrhofer, Benedikt

AU - Hagmueller, Martin

AU - Pernkopf, Franz

AU - Aichinger, Philipp

PY - 2025

Y1 - 2025

N2 - Pathological speech, caused by dysphonia or produced via electro-larynx devices, often suffers from poor intelligibility and unnatural prosody. In this paper, we investigate the potential of four state-of-the-art voice conversion models: FreeVC, QuickVC, LLVC, and XVC for restoring healthy-sounding speech. All models are fine- tuned on Austrian-German datasets and evaluated using objective and subjective metrics. Results show substantial gains in intelligibility, naturalness, and perceived vocal health. QuickVC, FreeVC, and XVC perform similarly and achieve the highest preference scores, exceeding unprocessed pathological speech by up to 200%. These findings highlight the potential to improve communication for individuals with voice disorders and motivate further development of efficient, high-quality conversion systems.

AB - Pathological speech, caused by dysphonia or produced via electro-larynx devices, often suffers from poor intelligibility and unnatural prosody. In this paper, we investigate the potential of four state-of-the-art voice conversion models: FreeVC, QuickVC, LLVC, and XVC for restoring healthy-sounding speech. All models are fine- tuned on Austrian-German datasets and evaluated using objective and subjective metrics. Results show substantial gains in intelligibility, naturalness, and perceived vocal health. QuickVC, FreeVC, and XVC perform similarly and achieve the highest preference scores, exceeding unprocessed pathological speech by up to 200%. These findings highlight the potential to improve communication for individuals with voice disorders and motivate further development of efficient, high-quality conversion systems.

M3 - Chapter

SN - 979-12-215-0820-8

SP - 33

EP - 36

BT - Models and Analysis of Vocal Emissions for Biomedical Applications

PB - Firenze University Press

T2 - 14th International Workshop on Models and Analysis of Vocal Emissions for Biomedical Applications, MAVEBA 2025

Y2 - 16 December 2025 through 17 December 2025

ER -