# Menschen German — Co-learning Context Repo

This repo tracks what I've actually covered in the *Menschen* German course
(course book + workbook), lesson by lesson, so that Claude (or any LLM) has
exact context on my vocabulary and grammar level when we study together.

## How to use this with Claude

At the start of a study session, point Claude at the relevant lesson file(s)
under the current book folder (e.g. `A1.1/course-book/lesson-05.md` and
`A1.1/workbook/lesson-05.md`), or the whole `A1.1/INDEX.md` for a running
summary. Ask Claude to only use vocabulary/grammar that appears in those
files (or earlier lessons) when creating exercises, example sentences, or
corrections — this keeps practice within what I've actually learned.

## Structure

```
A1.1/
  course-book/lesson-XX.md   ← vocab, grammar, dialogues from the course book
  workbook/lesson-XX.md      ← additional vocab/exercise types from the workbook
  INDEX.md                   ← one-line summary of every lesson so far
A1.2/                        ← added once I move on to the next book
  ...
```

## Adding a new lesson

1. Upload photos or a PDF of the course book / workbook pages for that lesson.
2. Claude extracts content into `lesson-XX.md` using `_template.md`.
3. Review and correct anything Claude got wrong or over-summarized.
4. Update `INDEX.md` with a one-line entry for the lesson.
5. Commit.

## Audio (listening exercises)

Hueber (the publisher) releases official **Transkriptionen** PDFs for both
the Kursbuch (course book) and Arbeitsbuch (workbook) — these are the
primary source for audio content, not the *Lösungsschlüssel* (answer key,
found at the back of the workbook, which is a different thing). Raw audio
files are **not** stored in this repo (see `.gitignore`); only transcripts
(text) are, since that's what gives Claude context.

Primary workflow — official transcripts:
1. Get the Kursbuch and Arbeitsbuch Transkriptionen PDFs for the relevant book
   (A1.1, A1.2, ...) — search "Menschen [level] Kursbuch/Arbeitsbuch
   Transkriptionen pdf Hueber", or check the publisher's site.
2. Send/upload the pages for a lesson the same way as course book/workbook
   content — Claude extracts the per-track dialogue text into that lesson's
   `.md` file under "Audio / Listening".
3. No review needed for accuracy — these are publisher-official text, not ASR.

Fallback workflow — local ASR (only if no official transcript exists for a
given track, e.g. supplementary listening material):
1. Get the audio files locally (publisher CD/app/download), any folder outside git.
2. Run `scripts/transcribe.py` (uses `faster-whisper` on GPU) to transcribe them:
   ```bash
   pip install faster-whisper
   python scripts/transcribe.py "<path-to-lesson-audio>" "A1.1/course-book/transcripts/lektion-XX"
   ```
3. Skim the `.txt` output for ASR mistakes (names, numbers, fast speech) and fix them.
4. Reference/paste the transcript into the lesson's `.md` file under "Audio / Listening",
   flagged as ASR-generated (the template already does this), or point at the
   `transcripts/lektion-XX/*.txt` path — transcripts are small text files so
   committing them is fine.

## Adding a new book (e.g. A1.2)

Copy the `A1.1/` folder structure (minus content) into a new `A1.2/` folder
and repeat the process.