Home Publications Research Grants Inventions & Patents Awards Additional Research Activities Faculties & Institutions Research Areas

Enhancing ASR Accuracy for Speakers with Parkinson’s Disease Using Instruction-Tuned LLMs

Hernandez A, Arias-Vergara T, Maier A, Perez Toro PA (2025)

---

Publication Type: Conference contribution

Publication year: 2025

Journal

Publisher: Springer Science and Business Media Deutschland GmbH

Book Volume: 16030 LNAI

Pages Range: 153-164

Conference Proceedings Title: Lecture Notes in Computer Science

Event location: Erlangen, DEU

ISBN: 9783032025500

DOI: 10.1007/978-3-032-02551-7_14

Abstract

Automatic speech recognition (ASR) systems often struggle with dysarthric speech, particularly from individuals with neurodegenerative conditions such as Parkinson’s disease. Addressing this challenge, we explore the use of instruction-tuned large language models (LLMs) for ASR error correction. Our approach uses zero-shot prompting with only the 1-best greedy ASR output. Using speech data from the Speech Accessibility Project, we fine-tune a 250M parameter Flan-T5 model to improve transcription accuracy for dysarthric speakers. Evaluation against raw outputs from multiple open-source ASR systems demonstrates notable improvements in correction quality. We further assess larger models, such as the 8B parameter LLaMA 3.1 model, and observe additional gains with minimal hyperparameter tuning. These results highlight the potential of LLM-based correction to enhance ASR accessibility for individuals with speech impairments.

Authors with CRIS profile

How to cite

APA:

Hernandez, A., Arias-Vergara, T., Maier, A., & Perez Toro, P.A. (2025). Enhancing ASR Accuracy for Speakers with Parkinson’s Disease Using Instruction-Tuned LLMs. In Kamil Ekštein, Miloslav Konopík, Ondrej Pražák, František Pártl (Eds.), Lecture Notes in Computer Science (pp. 153-164). Erlangen, DEU: Springer Science and Business Media Deutschland GmbH.

MLA:

Hernandez, Abner, et al. "Enhancing ASR Accuracy for Speakers with Parkinson’s Disease Using Instruction-Tuned LLMs." Proceedings of the 28th International Conference on Text, Speech, and Dialogue, TSD 2025, Erlangen, DEU Ed. Kamil Ekštein, Miloslav Konopík, Ondrej Pražák, František Pártl, Springer Science and Business Media Deutschland GmbH, 2025. 153-164.

BibTeX: Download