Skip to content

Home

Easily create audio for passive input

Condense anime audio into just the dialogue. Optionally include the OP and ED. No need to find or sync any subtitles.


Koemoe

Features

  • Generates condensed audio from video files without subtitle files
  • Optionally include the OP and ED in the final condensed file
  • Input a single video or a folder of videos
  • Uses ML model based on the YOHO paper with a convnext backbone

Planned Updates

  • Automatic Model Updates on Startup
  • Remove Fastai requirement for running the model
  • Options to change some of the model output filtering and smoothing such as the threshold and padding values

Caveats

  • The quality of the output can vary depending on the content. I will be adding options in the future for more aggressive or less aggresive filtering. However, the quality of the model should improve over time as I expand and modify the dataset.

Alternatives + Other Tools

Condensing

1. subs2cia - Condenses audio using subtitle files. I was using this before developing Koemoe.

Subtitle Retiming

1. auto-sub-retimer For retiming subtitles to videos. Very useful for syncing JP subs to JP content. The only cavaet is if the subtitles are image based, this will not work. There are some workarounds for this.