jpmschweitzerandClaude 67bee80dc8 docs(architecture): sync with measured 2026-08-07 state ...

Every figure in the latency budget was stale, in both directions. TTS was
listed at ~1.9 s per sentence but measures ~0.24 s warm for 4.5 s of audio;
the full Tatlock flow was listed at 11-25 s but measures ~10-13 s for simple
turns. Both sets of numbers predate the current model.

The VRAM section now carries real figures and the reason they matter: on
2026-08-07 Tatlock ran against a 9.3 GB model, leaving 7 MiB free, and every
transcription failed with CUDA out of memory while the Speaches container
still reported healthy. The budget is the constraint, not slack.

Also replaces the retired tatlock.schweitz.internal hostname in the topology
diagram with the docker container name.

Co-Authored-By: Claude <noreply@anthropic.com>

2026-08-07 15:05:15 +02:00

..

architecture.md

docs(architecture): sync with measured 2026-08-07 state

2026-08-07 15:05:15 +02:00

upstream-sdio-wedge.md

Upstream reproduction pack for the SDIO wedge (esp-hosted-mcu#167)

2026-07-15 02:12:11 +02:00