Have you tried the latest patch? We rev’d it to 1.5.1a to swap in some shorter voice samples that seemed to fix that issue