Quick start

• Clone from GitHub

  • git clone https://github.com/otherroch/anytran.git

• cd anytran

• Install system prerequisites

  • python, ffmpeg and other audio packages
  • sudo ./install_pkg.sh

• Create python 3.12 environment using venv

  • python3.12 -m venv .venv on Linux/Mac

  OR

  • py -3.12 -m venv .venv on Windows

• alternatively create python 3.12 environment using conda

  • conda create -n anytran python=3.12

    (specify a name and python version)

• Activate the environment

  • source .venv/bin/activate or . .venv/bin/activate on Linux/Mac

  • .venv\Scripts\activate on Windows

    OR

  • conda activate anytran

• Install all features

  • may require python -m pip install -U pip
  • pip install --group all -e .

• For individual feature installs, GPU support, or Linux system prerequisites, see the installation section

Run

Examples:

• RTSP:
  • anytran --rtsp rtsp://... --scribe-text transcript.txt
• System output (Windows):
  • anytran --from-output --scribe-text transcript.txt
• Web server:
  • anytran --web --scribe-text transcript.txt
• YouTube:
  • anytran --youtube-url https://... --youtube-api-key YOUR_KEY --scribe-text transcript.txt
• Translate to French (text):
  • anytran --rtsp rtsp://... --output-lang fr --slate-text french.txt
• Translate to French (voice):
  • anytran --from-output --output-lang fr --slate-voice
• File input (audio):
  • anytran --input sample.wav --scribe-text transcript.txt
• File input (audio -> translate):
  • anytran --input sample.wav --output-lang fr --slate-text french.txt
• File input (text -> translate):
  • anytran --input notes.txt --input-lang en --output-lang es --slate-text spanish.txt
• File input (text -> translate, local AI backend):
  • anytran --input notes.txt --input-lang en --output-lang fr --slate-text french.txt --slate-backend marianmt
• Generate voice table (for voice matching):
  • anytran --voice-table-gen --voice-table-lang fr,en

Timing summaries

• anytran --input sample.wav --timers # Print timing summary by stage
• anytran --input sample.wav --timers-all # Print all timing summaries (full, by stage, overhead)

Documentation

For a complete reference of all command line options and how they interact with each other, see:

• Command Line Options Reference

Additional topic guides:

• Installation — CUDA GPU support, individual feature installs, and Linux system prerequisites
• Text Translation Backends — details on all translation backends (googletrans, libretranslate, MarianMT, MetaNLLB, TranslateGemma, Gemma 4)
• Gemma 4 Setup — guide for using Google's Gemma 4 multimodal model for transcription and translation
• TranslateGemma Setup — guide for using Google's local TranslateGemma AI model
• TTS Backends — details on all voice synthesis backends (gtts, piper, custom/Qwen3-TTS, fish-speech, IndexTTS)
• Voice Matching — auto-matching voice features
• Loop Translation (looptran) — iterative back-translation with --looptran and --tran-converge

License

This project is licensed under the Apache License 2.0.