Benjamin Admin 19cbbf310a Improve garbled IPA cleanup: trailing strip, prefix match, broader guard ...

1. Strip trailing garbled IPA after proper [IPA] brackets
   (e.g. "sea [sˈiː] si:" → "sea [sˈiː]")
2. Add prefix matching for merged tokens where OCR joined headword
   with garbled IPA (e.g. "schoolbagsku:lbæg" → "schoolbag [skˈuːlbæɡ]")
3. Broaden guard to also trigger on trailing non-dictionary words
   (e.g. "scare skea" → "scare [skˈɛə]")

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

2026-03-19 10:36:25 +01:00

.claude

…

.gitea/workflows

…

.woodpecker

…

admin-lehrer

Add exclude region marking in Structure step

2026-03-19 09:08:30 +01:00

agent-core

…

backend-lehrer

…

breakpilot-drive

…

docs-src

feat: add border ghost filter + graphic detection tests + structure overlay

2026-03-16 18:28:53 +01:00

edu-search-service

…

klausur-service

Improve garbled IPA cleanup: trailing strip, prefix match, broader guard

2026-03-19 10:36:25 +01:00

paddleocr-service

fix: update PaddleOCR init for v3.4+ API (lang=en, ocr_version=PP-OCRv5)

2026-03-13 09:39:33 +01:00

school-service

…

scripts

chore: add mflux model download script for off-peak scheduling

2026-03-05 11:20:53 +01:00

studio-v2

feat: breite Spalten per Word-Gap splitten + gedrehte Scans im Frontend anzeigen

2026-03-07 18:16:32 +01:00

voice-service

refactor: Jitsi/Matrix/Voice von Core übernommen, Camunda/BPMN gelöscht, Kommunikation-Nav

2026-03-05 17:01:47 +01:00

website

refactor: LLM Compare komplett entfernt, Video/Voice/Alerts Sidebar hinzugefuegt

2026-03-05 17:34:54 +01:00

.env.example

chore: LLM qwen3:30b-a3b → qwen3.5:35b-a3b

2026-03-06 07:32:39 +01:00

.gitignore

…

docker-compose.yml

feat: PaddleOCR Remote-Engine (PP-OCRv5 Latin auf Hetzner x86_64)

2026-03-12 09:31:22 +01:00

mkdocs.yml

…

Description

No description provided

42 MiB

Languages

TypeScript 60.2%

Python 32.9%

Go 5.5%

C# 0.8%

CSS 0.2%

Other 0.3%