Skip to content

Install TATERS

Three steps. If you have ever installed any software in your life, you are overqualified.

1. Install Python

TATERS needs Python 3.10 or newer. If you are choosing today, take 3.13: two things have no Python 3.14 release yet — speaker diarization and training word vectors — so on 3.14, everything else installs and works, and those two are simply missing. If you already have 3.10–3.13, skip ahead.

  • Windows — grab the Python installer from python.org/downloads and run it. Tick the "Add Python to PATH" box on the first screen; that is the one checkbox that saves you a headache later.
  • macOS — grab the installer from python.org/downloads and run it. (If you use Homebrew, brew install python@3.13 works too.)
  • Linux — you almost certainly have it. If not: sudo apt-get install python3 python3-pip.

2. Install TATERS

If you already use Python for other things, glance at "I like keeping my Python installs tidy" in the notes below first. I strongly recommend that you run TATERS in a virtual environment (a "venv"). Trust me: a virtual environment takes one command now to get it set up and saves you about 10,000 headaches later. Really, just trust me.

But, if you don't want to use a virtual environment... go ahead and open a terminal (Windows: search for "PowerShell" or "cmd"; macOS: search for "Terminal") and type:

pip install taters

That's it. Pip fetches TATERS and everything it needs.

3. Run it

One installed: in that same terminal, type:

taters

Voilà — the TATERS app opens right there in your terminal and walks you through everything else: what your data is, what you want out of it, and where the results should go. You never have to write a line of code.

Optional notes (skip these unless they describe you)

I have an NVIDIA GPU and want to use it

Do these before step 2, in this order:

  1. Install a current NVIDIA driver (and CUDA support) from nvidia.com/drivers.

  2. Install the CUDA 12.8 build of PyTorch:

    pip install torch torchvision --index-url https://download.pytorch.org/whl/cu128

  3. Then pip install taters.

Use cu128, not the newest CUDA your driver allows. PyTorch's own installer page will offer you a newer one, and transcription cannot use it. cu128 covers every current card, so there is nothing to gain by going higher — Installation troubleshooting explains why.

Already installed TATERS first? No problem — inside the app, go to Check my setup and it will tell you exactly what to run to switch PyTorch onto your GPU.

I'll be working with audio or video files

TATERS uses FFmpeg to read media files — it is the one thing pip can't install for you.

  • Windows: winget install Gyan.FFmpeg (then close and reopen the terminal)
  • macOS: brew install ffmpeg
  • Linux: sudo apt-get install ffmpeg

Working with text only? You don't need this.

I like keeping my Python installs tidy

A virtual environment keeps TATERS and its dependencies from mingling with your other Python projects — like keeping the mashed potatoes out of the fruit salad:

python -m venv venv-taters
# Windows:  venv-taters\Scripts\activate
# macOS/Linux:  source venv-taters/bin/activate
pip install taters

TATERS now lives in your virtual environment named venv-taters and is kept isolated from the rest of your Python environment. This helps make sure that TATERS doesn't overwrite anything that might be in use by your other packages. Just remember to activate your venv again whenever you open a new terminal.

Something not working? See installation troubleshooting, or run TATERS and pick Check my setup — it inspects your machine and tells you, in plain words, what (if anything) needs fixing.