feat(voice): local-only LLM + faster turns #166

2026-06-17T17:49:45Z

electron commented

2026-06-17 17:49:45 +00:00

Local-only NPC LLM (granite, no remote), persona KV-cache prewarm at greeting, temperature+repetition_penalty (no looping), STT pad 2s->0.5s, submodule bump (hook debounce cfe429d).

electron added 1 commit 2026-06-17 17:49:46 +00:00

feat(voice): local-only LLM + faster turns

Repo State / repo-state (pull_request) Failing after 43s

Details

Validate Zacus refactor / validate (pull_request) Failing after 6m35s

Details

Repo State / repo-state (push) Failing after 45s

Details

Validate Zacus refactor / validate (push) Failing after 13m52s

Details

643c3852de

Gateway: drop the remote ailiance fallback — the NPC LLM is now the
on-box granite ONLY (local spoken line on failure, never remote).
Prewarm the persona KV-cache in the background at greeting time so the
first reply is fast instead of paying ~17 s of cold prompt processing.
Add temperature + repetition_penalty so granite4-tiny stops looping.
STT: cut the trailing-silence pad 2 s -> 0.5 s (callers already pad),
shaving STT latency. Bump ESP32_ZACUS to cfe429d (hook debounce).

electron merged commit 643c3852de into main

2026-06-17 17:50:24 +00:00

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: electron-rare/le-mystere-professeur-zacus#166