https://github.com/jeeftor/chapterize-whisper
Chapterizer using Whisper
https://github.com/jeeftor/chapterize-whisper
Last synced: 7 months ago
JSON representation
Chapterizer using Whisper
- Host: GitHub
- URL: https://github.com/jeeftor/chapterize-whisper
- Owner: jeeftor
- Created: 2024-12-20T20:35:32.000Z (over 1 year ago)
- Default Branch: master
- Last Pushed: 2024-12-27T02:21:01.000Z (over 1 year ago)
- Last Synced: 2024-12-27T03:21:44.725Z (over 1 year ago)
- Language: Python
- Size: 990 KB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Chapterize-Whisper
_A utility written to help get better chapters for Audiobookshelf_
Uses whisper and will make some automatic stuff
Batch mode is faster - but - it seems to give worse results for chapterfinding - so we use a slower option.
To install globally I think you can use:
```bash
pip install -U chapterize-whisper
chapterize --help
```
# Example
Download a free audiobook like this one:
https://librivox.org/the-adventures-of-danny-meadow-mouse-by-thornton-w-burgess/
and put it in some directory. Then we can use the `chapterize` command like:
```bash
chapterize detect --dir ./audio/danny_meadow_mouse_1301_librivox
```
It will run a transcription process looking like this maybe:
Once done you'll get a `.chapter` file in the directory of your audio files. For example this is what came out:
```
00:00:00,0000, BOOK Start
00:01:05,920, Chapter 1, Danny Meadow Mouse is worried.
00:04:48,160, Chapter two, Danny Meadow Mouse and his short tail.
00:08:06,120, Chapter 3, Danny Meadow Mouse plays hide and seek.
00:13:04,759, Chapter four, old Granny Fox tries for Danny Meadow Mouse.
00:17:17,137, Chapter 5.
00:26:13,777, Chapter seven, old Granny Fox tries a new plan.
00:30:15,497, Chapter 8 Brother Northwind proves a friend.
00:35:00,017, chapter 9, Danny Meadowmouse's caught at last.
00:39:38,975, CHAPTER X
00:44:43,015, Chapter 11, Peter Rabbit gets a fright.
00:48:35,415, Chapter 12, the old briar patch has a new tenet.
00:52:44,135, Chapter 13.
00:56:40,975, Chapter 14, Farmer Brown sets a trap.
01:01:25,957, Chapter 15 Peter Rabbit is caught in a snare.
01:05:32,077, Chapter 16 Peter Rabbit's hard journey.
01:07:05,957, Part of the stake to which the snare had been fastened and which Peter had managed to
01:10:10,397, Chapter 17 Danny meadow mouse becomes worried.
01:15:02,917, Chapter 18, Danny meadow mouse returns a kindness.
01:19:15,077, Chapter 19, Peter Rabbit and Danny Meadowmouse live high.
01:23:41,237, Chapter 20.
01:28:46,237, Chapter 21, an exciting day for Danny Meadow Mouse.
01:33:23,997, Chapter 22
01:41:31,997, Chapter 24
01:46:18,820, BOOK_END
```
You'll notice some chapters are missing a full description - and some are wrong such as:
`00:27:43,702, Part of the stake to which the snare had been fastened and which Peter had managed to`
For example to fix `Chapter 5.` open up the `.srt` file and look for where chapter 5 is:
```srt
318
00:17:17,137 --> 00:17:18,977
Chapter 5.
319
00:17:18,977 --> 00:17:22,137
What Happened on the Green Meadows
```
In this case we can assume the correct title is:
`Chapter 5. What Happened on the Green Meadows`
Also chapter 6 is missing - you can see in the srt it is here:
```srt
373
00:21:39,017 --> 00:21:48,497
End of Chapter 5 Chapter 6 Danny Meadow Mouse remembers, and ready Fox, poor kids.
```
This chapter will likely take some tweaking since you need to figure out how long the "End of Chapter 5" takes ... but you can also just guess - lets say 2 seconds. So you'd add this to the chapter file:
```
00:21:41, Chapter 6 Danny Meadow Mouse remembers
```
Once you've cleaned up the chapter files it will look like something like this. Maybe you are too lazy and don't want to fill out all the chapter titles - thats fine.
```
00:00:00,0000, BOOK Start
00:01:05,920, Chapter 1, Danny Meadow Mouse is worried.
00:04:48,160, Chapter two, Danny Meadow Mouse and his short tail.
00:08:06,120, Chapter 3, Danny Meadow Mouse plays hide and seek.
00:13:04,759, Chapter four, old Granny Fox tries for Danny Meadow Mouse.
00:17:17,137, Chapter 5.
00:21:40,000, Chapter 6.
00:26:13,777, Chapter 7, old Granny Fox tries a new plan.
00:30:15,497, Chapter 8 Brother Northwind proves a friend.
00:35:00,017, chapter 9, Danny Meadowmouse's caught at last.
00:39:38,975, CHAPTER 10
00:44:43,015, Chapter 11, Peter Rabbit gets a fright.
00:48:35,415, Chapter 12, the old briar patch has a new tenet.
00:52:44,135, Chapter 13.
00:56:40,975, Chapter 14, Farmer Brown sets a trap.
01:01:25,957, Chapter 15 Peter Rabbit is caught in a snare.
01:05:32,077, Chapter 16 Peter Rabbit's hard journey.
01:10:10,397, Chapter 17 Danny meadow mouse becomes worried.
01:15:02,917, Chapter 18, Danny meadow mouse returns a kindness.
01:19:15,077, Chapter 19, Peter Rabbit and Danny Meadowmouse live high.
01:23:41,237, Chapter 20.
01:28:46,237, Chapter 21, an exciting day for Danny Meadow Mouse.
01:33:23,997, Chapter 22
01:41:31,997, Chapter 24
01:46:18,820, BOOK_END
```
The next step would be to upload this "chapter definition" to audiobook shelf.
First we need to upload this audiobook itself to Audiobookshelf - and as you can see - it doesnt have any chapter data
If you look at Audiobookshelf's URL you'll see something like this:
`https://audiobookshelf.local/item/2b1a5c2f-02e4-47bb-99ab-cc800aeafec7`
What we care about is the UUID for the item:
`2b1a5c2f-02e4-47bb-99ab-cc800aeafec7`
So with this information we can do the following:
```bash
chapterize upload \
--dir ./audio/danny_meadow_mouse_1301_librivox \
--id 2b1a5c2f-02e4-47bb-99ab-cc800aeafec7 \
--abs-url https://audiobookshelf.local \
--api-key this-is-a-fake-api-key
```
[//]: # (### Uploading chapter data)
[//]: # ()
[//]: # (Next we will use the API https://api.audiobookshelf.org/#book-chapter to turn our .chapter_xxx file into an array of `book chapters` which can be uploaded to audiobookshelf)
[//]: # ()
[//]: # (and when we are done it looks like this:)
