Skip to content

LiveTL/kanatran

Repository files navigation

Kanatran

An automatic livestream caption/translation system for Japanese Vtubers [WORK IN PROGRESS]

Build Runner Build Watcher Deploy to Production

About

LiveTL is building an automatic livestream caption/translation system for Japanese Vtubers, and we need your support.

I'll get straight to the point: we developed a system called Kanatran to generate machine translations for Japanese Vtuber streams in realtime, and we need some funding to deploy it. Below are the things we want to achieve:

  1. Run Kantran on Japanese Hololive and Holostars streams, as well as selected agencies / independents
  2. Provide Kanatran translations to users of LiveTL
  3. Make our automated translations accessible to other 3rd party apps and websites for watching Vtubers

To do this, we need some financial support to kickstart our efforts. Our team accepts donations through Open Collective, a platform that makes our spendings completely transparent to donators. We will use funding from supporters to run our servers and translation software. We currently have about $750, which would only last a few short months. Please consider chipping in here: https://opencollective.com/livetl

Our system is extremely low cost compared to previous attempts at machine translations, and we will make our entire transcription/translation technology completely free and open source here on GitHub. Additionally, we will make our machine translations as accessible as possible -- if you are developer that wants to access our translations, please reach out to us (preferably through our Discord server). Our goal is not to monopolize this technology, but to use it to improve the Vtuber-watching experience for everyone in this community.

Usage

  1. Pull the latest images

    docker pull ghcr.io/livetl/watcher:latest
    docker pull ghcr.io/livetl/runner:latest
  2. Create a bridge network

    docker network create -d bridge kanatran
  3. Configure a .env file

    LIVETL_API_KEY=
    INTERCOM_NETWORK=kanatran
    MAX_CPU=50%
    MAX_MEM=4GB

    View All Configurable Environment Variables

  4. Run the pre-built image

    docker run -v /var/run/docker.sock:/var/run/docker.sock --env-file .env ghcr.io/livetl/runner

Environment Variables

Variable Description Required Values Default
LIVETL_API_KEY LiveTL API Key String
WATCHER_IMAGE Watcher Image Name String ghcr.io/livetl/watcher
API_URL API URL String https://api.livetl.app
INTERCOM_NETWORK Inter-container Bridge Network Name String kanatran
CONTROLLER_URL Controller Address String wss://api.livetl.app/kanatran/controller
INTERCOM_PORT Inter-container Bridge Port Integer 6969
MAX_CPU Max CPU Usage Percentage (__%) 100%
MAX_MEM Max Memory Usage Bytes (__GB, __MB, etc.), Percentage (__%) 100%

Development

Requirements

  • docker
  • docker-compose
  • make
  • npm
  • node
  • python

Setup

git clone https://github.com/LiveTL/Kanatran

Dev Info

Commands

  • make init: Initialize the controller server
  • make build image={runner or watcher}: Build images
  • make start: Start the controller server
  • make run: Run the local runner
  • make spawn video={video id}: Start a translation container

Environment Variables

  • MODE: production or development for the controller