dubAI

AI-powered dubbing toolset that I am working on. It's mostly for my own usage but I try to make it appropiate enough for other people to use, too.

TODO

• Generate SRT

• Translate SRT [partially done]

• Dub according to SRT [wip]

  • Get SRT fragment to know timing, position and text
  • Create n audio files according to the SRT timings
  • Send i audio file together with its corresponding text to the dubber LLM, n times
  • Save the files
  • All of this in a proper folder

• Mix original audio with dubbed audio

  • Duck original audio's volume according to the SRT time fragments
  • Adjust audio as needed

  Note: I have to have koboldcpp running before the voice references are created! Also, decide whether I will

• Pass LLM addresses thru CLI

Principles

• Modular

  You should be able to use any tool you want for each step of the process, if possible.

TODOs

• Implement own error types