NeuroSpeech

NeuroSpeech models and analyzes pathological speech signals to quantify phonation, articulation, prosody, intelligibility, and to enable discrimination and clinical-scale prediction for neurological conditions such as Parkinson's disease.

Key Features:

• Multidimensional speech modeling: Models phonation, articulation, prosody, and intelligibility from recorded speech signals.
• Automatic discrimination: Automatically distinguishes Parkinson's versus non-Parkinson's speech signals.
• Neurological state prediction: Predicts Unified Parkinson's Disease Rating Scale (UPDRS) scores from speech-derived features.
• Dysarthria level assessment: Predicts dysarthria severity mapped to the Frenchay Dysarthria Assessment scale.
• Adaptability: Provides a framework that can be extended to other neurological pathologies and additional measures.

Scientific Applications:

• Clinical diagnosis: Supports diagnosis of Parkinson's disease and assessment of speech impairment severity using UPDRS and Frenchay-derived predictions.
• Research: Enables analysis of pathological speech characteristics across phonation, articulation, prosody, and intelligibility dimensions.
• Extension to other conditions: Facilitates adaptation to study other neurological disorders and to incorporate new evaluation measures.

Methodology:

Performs automatic discrimination between Parkinson's and non-Parkinson's speech and predicts UPDRS scores and Frenchay Dysarthria Assessment levels from speech-derived features.