Skip to content

Latest commit

 

History

History
53 lines (39 loc) · 1.3 KB

README.md

File metadata and controls

53 lines (39 loc) · 1.3 KB
Raw

KTV

Split vocal and accompaniment for a given youtube URL.

Requirements

Download Video From Youtube and Process Audio Separation

For example, to split this song and create a processed video

docker run \
  -e MODEL_PATH=/tmp/ktv/models \
  -v /tmp:/tmp \
  -t adason/ktv:latest \
  -o luggage \
  https://www.youtube.com/watch?v=51SOZjKuaaQ

The processed output video will include 3 audio tracks:

  • First audio track from the original source
  • Second audio track for accompaniment only
  • Third audio track for vocals only

You can find processed files such as luggage_processed.mp4 under /tmp/ktv/ along with other intermediate files with prefix luggage. You can change docker mount point to any local directory (such as -v $HOME/Downloads:/tmp). You can also replace the filename output prefix luggage as desired.

Development

Docker

docker build -t ktv .

docker run \
  -e MODEL_PATH=/tmp/ktv/models \
  -v /tmp:/tmp \
  -t ktv \
  -o luggage \
  https://www.youtube.com/watch?v=51SOZjKuaaQ

Push to docker registry

docker tag ktv adason/ktv:latest
docker push adason/ktv:latest

Python virtualenv

python -m ktv -o luggage https://www.youtube.com/watch?v=51SOZjKuaaQ