Merge the MicroslopSam engine into the repository

This commit is contained in:
Fisto
2026-09-21 12:16:33 +02:00
parent 51d42b3a16
commit bbb580deea
30 changed files with 8959 additions and 82 deletions
+9
View File
@@ -40,6 +40,15 @@ mobile draft at 390px.
- [x] C11 preview caching: Discord-hosted image/GIF previews (WebP 320px, hash-addressed, 5 MiB
cap, one attempt) run after every walk in both modes. Live run `...-20260916090722`:
`previewed: 2`; thumbs served over `/api/media/<scrape>/<item>/thumb` (200).
- [x] C66 Merged the SAM engine into Veskslop. `microslopsam/` is now a full copy of the
standalone project as it ran (wrapper, docs, README, build artifacts, and the vendored
binaries SAM tracks), the runtime env points at `microslopsam/setup/veskslop-tts.sh` with
regenerated voices, and `microslopsam/setup/microslopsam.env` resolves inside the repo. The
word-mode + preset fix from C65 landed on the merged wrapper (word mode wins over VOX prosody,
parser skipped, per-word rate/pitch tags, preset-aware `--voices`); the wrapper test harness
passes and a real-engine run shows 9 per-word gaps at 11.1 s versus 7.5 s for a single
utterance. An `env -i` run proved the copy runs without the SAM project. Process rule added to
AGENTS.md: read TTS-DESIGN first, keep the in-repo copy current, run the wrapper tests.
- [x] C65 Word mode with the Half-Life preset. Word mode no longer yields to VOX prosody: with
a preset it synthesizes each word with the preset profile (per-word `\!R`/`\!%` tags), and the
VOX parser is skipped because its tags cannot survive a word split. The daemon always exports