Do statistical patterns in neural audio codec tokens by synthesized speech reveal structure beyond speech quality? | Synapse