Measure CrisperWhisper 2.0 turbo in Russian before wiring a runtime for it #207

2026-08-08T22:29:12+02:00

claude commented

2026-08-08 22:29:12 +02:00

Closes Vikunja #665.

Acceptance criteria and quality gate are in TASK.md on this branch.
Review the review, not the diff — leave comments and the agent will apply them via task start 665.

Closes Vikunja #665. Acceptance criteria and quality gate are in `TASK.md` on this branch. Review the review, not the diff — leave comments and the agent will apply them via `task start 665`.

claude added 1 commit 2026-08-08 22:29:12 +02:00

Measure CrisperWhisper 2.0 in Russian against the deployed floor (V-665) 4666057066

Turbo in Intended mode scores 10.4% WER on 200 Golos crowd clips, against
27.5% for the ggml-small.bin the box loads today. It also beats its own base
model and CW2 large, which inverts what the card implies about turbo.

The mode choice is not settled by this corpus. Intended and verbatim disagree
on 29 of 200 after normalization, and the disagreement is script rather than
disfluency. Golos crowd carries almost no disfluency to disagree about.

whisper.cpp cannot load CW2: num_languages() derives from n_vocab and CW2's
51897 shifts seven special token ids. So the runtime is workpc under V-486,
with whisper on homesrv as the floor.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013ptwopxyo3Z2kwFckHkLvN

claude merged commit 672eabc134 into master

2026-08-08 23:21:14 +02:00

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: kami/Maven#207