Autocomper

Autocomper is a GUI frontend for sound_reader.py: a useful script that takes in videos, extracts certain portions from them, and writes the result to a new video.

Using the Program

1. Install a release or build and run the program for your platform.

2. Click "Add Videos", then select any number of videos.

   • If you are combining the videos, you can change the order of their appearance in the final video by selecting a video in the list, then using the up/down arrows.
   • You can remove unwanted videos by selecting one or multiple videos from the list, then hitting "Remove Selected". You can also hit "Clear All" to remove all the videos at once.

3. Set the precision, block size, and threshold, and turn on any miscellaneous options. You can hover over each of the boxes to get more information about each parameter.

4. Click the "Select Output File" button in the top right to pick a place to write the final video(s).

   • If you selected "Combine Input Videos", you will select one specific file to write everything to. Otherwise, you will pick a directory, and individual videos will be written as {original_title}_comped.mp4.

5. Click "Process Videos" to begin the comping process. Depending on your input, this may take a while, especially when writing the final output video if you selected "Combine Input Videos".

6. Once the completion popup appears, navigate to your selected output location and find your video(s).

Building

Python Version: 3.10.11+

Windows

Ensure that Python is installed, then open a Powershell window at the root directory and run the following commands (or simply run the build_windows.ps1 script).

$ python -m venv .env
$ .\.env\Scripts\Activate.ps1
$ pip install -r requirements.txt
$ python setup.py build

The executable is written to build/exe.win-.../autocomper.exe.

Linux

Largely the same as Windows. First, ensure that python3 is installed:

$ sudo apt-get update
$ sudo apt-get install python3

Then, open a terminal window at the root directory and run the following commands.

$ python -m venv .env
$ source .env/bin/activate
$ pip install -r requirements.txt
$ python setup.py build

You can also simply run:

$ ./build_linux

The executable is written to build/exe.linux-.../autocomper. If permission is denied, please run the following command from the root directory.

$ chmod +x build/exe.linux-.../autocomper

OSX

No Mac build yet. Sorry :(

TODO

• Figure out how to build the program with onnxruntime-gpu instead of having to use boring old onnxruntime.
• Add other filetype options for output besides mp4.
• Add multiprocessing to video writing + switch to using raw FFMPEG instead of moviepy for maximum speed
• Add a "model download center" to avoid having to pack models in with the releases
• Add "open output" button on completion
• Add support for more input types (e.g. FLV)