Transcribe documentation¶
Transcribe is a local-first workbench for handwritten notebooks. You import scans you already have, transcribe them on your machine, and keep the results.
OCR uses a local Ollama vision model. Transcribe does not send pages to a cloud OCR service.
See how it works: user guide — import scans, transcribe, review beside the page.
Everyday jobs: import and transcribe, review, analyse, detect, export — user guide.
Is this for me? What Transcribe is.
Privacy: files stay on your computer; Ollama is local and off the public internet by default.
The GitHub README is the same first-run story.
Start here
Using Transcribe
Advanced
Developers
Contracts
- Analysis batch runs
- Analysis document
- Analysis result
- Analysis run storage
- Corpus integrity
- Detection definition
- Detection finding
- Detection result
- Detection run storage
- Fine-tune export
- Import runs
- Interface menus
- Notebook corpus
- Notebook eligibility
- Notebook export
- OCR batch runs
- OCR multipass
- OCR preference ledger
- Page metrics (ink / blankness / hue)
- Page results and provenance
- Project on-disk format
- Prompt definition
- Source assets
- Organisation tag catalog
- Workspace backup
- Workspace settings
Maintainer notes
- Contributing (docs and code orientation)
- Analysis module porting (from TranscriptX)
- Analysis port pins
- Analyse visual compare (TranscriptX alignment)
- Dependency audit
- Documentation architecture
- OCR Review workbench
- Places / NER map alignment (TranscriptX ↔ Transcribe)
- Release governance (manual next-tag checklist)
- Read the Docs go-live checklist
- Settings / profiles / models alignment (TranscriptX ↔ Transcribe)
- 0.9-1 unfamiliar user testing
Full lists¶
Usability wave (active product focus)