# Changelog

## Score

- CAI 60 → 58 (-2.8)
- Rubric changed (rubric-2026.10.1 → rubric-2026.10.5) — scores are not directly comparable.

## Lenses

- Code Health 85 → 89 (+3.6)
- Architecture 100 → 100 (+0.0)
- Maturity 53 → 46 (-6.9)
- Readiness 52 → 52 (+0.0)
- Security 75 → 75 (+0.0)

## Resolved (1)

- Workflow token permissions not restricted

## New (6)

- Hotspot: faster_whisper/transcribe.py (faster_whisper/transcribe.py)
- Orphaned file (benchmark/evaluate_yt_commons.py)
- Orphaned file (benchmark/memory_benchmark.py)
- Orphaned file (faster_whisper/feature_extractor.py)
- Orphaned file (faster_whisper/utils.py)
- Workflow token permissions not restricted (.github/workflows/ci.yml)

## Changes since last survey

- 9 commits — 8 feature/other, 1 fixes

## By area

- faster_whisper/transcribe.py — 5 commits
- faster_whisper/vad.py — 2 commits
- (root) — 1 commit
- faster_whisper/tokenizer.py — 1 commit

## Notable commits

- fix: Handle empty alignments in find_alignment (companion to the CTranslate2 divide-by-zero fix for #1342) (#1460)
- change: Accept a dict for vad_parameters in detect_language (#1662)
- change: Raise an error for illegal values in clip_timestamps arg (#1654)
- change: Raise an error for languages unsupported by a model (#1564)
- change: Tweak collect_chunks() (#1653)
- change: Update export LD_LIBRARY_PATH cmd in README.md (#1459)
- change: Update the initial_prompt parameter type hint in batched pipeline (#1299)
- change: Verify that config.json has the required keys (#1673)
- change: change language_detection_threshold type to float (#1188)