Abstract

Japanese anime voices exhibit distinctive expressive characteristics shaped by the culture of professional voice actors (seiyus). Prior research has focused mainly on global phonetic properties, leaving local, dynamic voice quality changes underexplored. This study analyzes non-modal voice qualities—breathiness, harshness, pressed voice, and inhalations— in performances by professional seiyus and seiyu-training students across dubbing, repeating, and reading tasks. Professionals showed a high rate of non-modal segments, whereas students and reading-style speech produced fewer, confirming that local voice modulation is central to anime performance. Breathiness appeared mainly at phrase-final positions, while harsh and pressed qualities were more common phrase-initially and medially. Acoustic measures (HNR, F1F3syn, H1–A1) reliably distinguished the voice quality categories and were consistent across speaker groups. Overall, the findings highlight local voice quality modulation as a key expressive feature of anime voice acting.