https://github.com/gregistech/lazyshorts-py

Create short videos, like a lazy person.
https://github.com/gregistech/lazyshorts-py

ai ffmpeg instagram lazy mediapipe moviepy shorts tiktok video videos whisper youtube

Last synced: 3 months ago
JSON representation

Create short videos, like a lazy person.

• Host: GitHub
• URL: https://github.com/gregistech/lazyshorts-py
• Owner: gregistech
• License: lgpl-3.0
• Archived: true
• Created: 2022-12-25T09:53:59.000Z (almost 2 years ago)
• Default Branch: master
• Last Pushed: 2024-04-09T07:34:42.000Z (7 months ago)
• Last Synced: 2024-08-03T13:05:36.491Z (3 months ago)
• Topics: ai, ffmpeg, instagram, lazy, mediapipe, moviepy, shorts, tiktok, video, videos, whisper, youtube
• Language: Python
• Homepage:
• Size: 69.3 KB
• Stars: 49
• Watchers: 2
• Forks: 9
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE.md

Awesome Lists containing this project

README

        
See [lazysys](https://github.com/lazysys) for the revamped version.

---

# LazyShorts

A command-line tool to convert long-form videos into multiple short-form videos, with burned-in text and subtitles.

It also cuts out unwanted silence.

## Preview

### Original video

https://user-images.githubusercontent.com/29675001/210073372-f2a3fa63-29c2-4bc0-ab03-31188a772021.mp4

### Result (with manual subtitle correction)

https://user-images.githubusercontent.com/29675001/210073848-09bcd7b7-6a86-4114-ab0b-72ddf674e298.mp4

### Result (without manual subtitle correction)

As you can see, in Hungarian the `medium` model works quite well: considering the bad quality of my input. 

The `large` model could be even better: **if** you have the hardware. :)

https://user-images.githubusercontent.com/29675001/210073868-f650cef7-093c-4260-9407-c6bc7de1b59e.mp4

## Notes

### Arguments

See `lazyshorts -h`

### Subtitles

I use Whisper to transcribe audible voices to text. 

Obviously with non-english languages the accuracy can be lower: you can help that by...

- ...using a different (for now, only Whisper) model (be wary, the `medium` model is hard to run even with 8GBs of RAM.)

- ...editing subtitles manually from segment to segment. (`{lazyshorts-py} e1 2 45 78`...)

### Not tested

- I don't know if running Whisper on GPU works, you could try CUDA. See `--whisper_device` and PyTorch/Whisper documentation. Also, get the CUDA enabled PyTorch as I define the CPU one in the `requirements.txt`.

### Known issues

- 'subprocess.run' somehow blocks UI process.

- We could use `rich` to have nice progress bars, as currently you have to manually poll the status of the renders.

- Cropping is just arbitrary: I wanted to use MediaPipe. It's not easy to even get it to run, but my resources were not enough. Maybe a less demanding model or cloud is needed?

- Don't combine segments that are less than `end_time`, you'll get an exception.