- Shipped
- October 4, 2026 at 10:56 PM UTC
- Author
- Kamo
- Commit
- 2cf2bc4
The speech model is sent the caller's input with every whole-word mention of the org's name or short name replaced by its phonetic spelling (OrgPhoneticSpeech). The org is read once per request, after the request is validated and before the walk, so every candidate is sent the same text, on both of SP96's paths (a person's or caller's voice, and an AI member's own voice). The 4096-character cap and the CHARACTERS_1K measure stay on the caller's input. An org that is not found, or cannot be read, is spoken as written; a failed read is a WARN.