Pandrator Editorial audio
PL Download

Quick start

Install only the components your first workflow needs.

The desktop launcher manages the application and optional local services. Docker and WSL are not required for the packaged workflow.

Open GitHub Releases

Install on Windows

  1. 1

    Download the installer

    Download PandratorInstaller.exe from GitHub Releases and run it.

  2. 2

    Choose a workspace and services

    Keep the default location or choose another one. Install only the speech or transcription services you need.

  3. 3

    Launch Pandrator

    The launcher starts the web application, worker, and selected services, waits for readiness, and opens the browser interface.

Reasonable first component sets

Reasonable first component sets

Lightweight audiobook

  • Kokoro
  • No LLM required
  • CPU is sufficient

Voice cloning

  • Qwen3 TTS Base or another compatible model
  • A clean reference recording
  • GPU optional; performance varies

Subtitles

  • CrispASR
  • Whisper for broad coverage or Parakeet for its supported languages
  • Speech service only if dubbing is needed

Local assisted workflow

  • A speech or transcription service
  • A local OpenAI-compatible LLM server
  • Enough memory headroom for concurrent models
Check language and hardware in the matrix

A few practical requirements

  • The packaged workflow does not require Docker or WSL.
  • Model downloads and caches can be large; check free disk space before selecting several services.
  • Calibre is optional and needed only for MOBI conversion.
  • Model licences differ and are shown by the installer before installation.