ABChapterize manual#
This is the complete reference for ABChapterize. For a quick start, see the README.
Download
Prebuilt binaries for Windows and Linux are on GitHub - no build tools, no runtime to install first. Unpack it and run it.
These pages document v1.0.0. They are published with each release, so they describe that version rather than whatever has been committed since.
Contents#
- 1. What ABChapterize does
- 2. Supported file formats
- 3. How detection works
- Analyze — silences, speech and music
- Probe — short windows where a chapter could start
- Re-probe — a cheap second try at a gap (only with a heavier --upgrade-model)
- Scan — reading a gap in full (only when needed)
- Re-scan — the shifted second reading
- Prologue and epilogue
- Custom marks
- Named marks that land beside a chapter
- Detecting chapters without believing their numbers
- Books that count from one again in every part
- The intro chapter
- 4. How chapters are written — file safety
- 5. What is kept and what is stripped
- 6. Command line reference
- Options from a file
- Target selection
- Audiobookshelf
- Detection behaviour
- Phrases and titles
- Auto language detection
- Handling of pre-existing chapters
- Detection safety nets
- Safety and undo
- Running your own commands around each file
- Cleaning up after a run
- Resuming an interrupted run
- Output
- Chapter export/import
- Performance
- Tuning constants
- Miscellaneous
- 7. Languages and number recognition
- 8. Whisper models
- 9. GPU acceleration
- 10. ffmpeg: requirements and discovery
- 11. xHE-AAC (USAC) files
- 12. Output, progress and logging
- 13. Exit codes
- 14. Troubleshooting