Introducing EMOPARKNZ: the Emotional Speech Database from New Zealand English Speakers with Parkinson’s Disease

This paper introduces EMOPARKNZ, a new speech database designed to study emotional speech production in New Zealand English speakers with Parkinson's disease. The database features 13 speakers (7 male, 6 female) and 1,950 speech recordings expressing five primary emotions: excited, angry, happy, sad, and neutral. The design considerations, participant selection, and recording procedures are detailed. Acoustic analysis of the database revealed significant differences in fundamental frequency, intensity, and speech rate across emotional states. A preliminary perception test with 22 New Zealand English listeners achieved 63% classification accuracy across the five emotions, aligning with the performance reported for similar emotional speech databases. The results indicate EMOPARKNZ is a valuable resource for advancing research in emotion communication for individuals with Parkinson's disease and speech emotion classification.