NeuroVoz: a Castillian Spanish corpus of parkinsonian speech

2024-03-04

Scientific Data

AI Summary

Current Parkinson’s disease (PD) voice screening research is severely hindered by the scarcity of publicly available non-English speech datasets—particularly Castilian Spanish—impeding methodological reproducibility and cross-linguistic validation. To address this gap, we present the first open-access Castilian Spanish speech corpus specifically designed for PD assessment, comprising 12 hours of high-fidelity recordings from 127 participants (62 PD patients, 65 healthy controls). Recordings were acquired using professional microphone arrays, aligned with UPDRS clinical scores, processed via Praat-based acoustic analysis, and annotated per ISO/IEC 23053 metadata standards. Crucially, we introduce a clinically interpretable acoustic–motor co-annotation framework to systematically characterize multidimensional vocal impairments. The dataset is publicly released and has already enabled three independent validation studies on PD voice biomarkers, thereby filling a critical void in non-English neurospeech resources.

Technology Category

• Natural Language Processing: Speech
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Cognitive Modeling & Cognitive Systems: Affective Computing

Application Category

• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Problem

• Lack of Parkinson's speech datasets in multiple languages
• NeuroVoz corpus addresses Castilian-Spanish Parkinsonian speech analysis
• Challenges in language-agnostic Parkinsonian speech pattern analysis

Innovation

• Created Castilian-Spanish Parkinson's speech corpus
• Included diverse speech tasks and expert assessments
• Achieved 89% accuracy in Parkinson's screening