Sing-song: a speakable encoding for long numbers and keys
Summary
Sing-song proposes a reversible, speakable encoding for arbitrary byte strings using a 64-syllable CV grammar that maps 6 bits per syllable. The post documents the encoding's properties, variant mechanism, test vectors, and a reference Python implementation, with potential use for human-friendly identifiers such as Nostr usernames. It emphasizes prefix stability, determinism, and the trade-off between written density and spoken readability.