COMMAND LINE

Atten, from your terminal.

Generate natural speech locally from scripts, automations, or a single command. No account or API key.

QUICK START

Set up once. Speak locally.

You need Git and uv. Atten uses its own pinned Python 3.12 environment, so it will not modify your system packages.

First run: the model and selected voice are downloaded once. After they are cached, generation works offline.

git clone https://github.com/jashdubal/atten.git
cd atten
uv python install 3.12
uv sync --frozen
uv run cli.py "Hello from Atten"

macOS · Windows PowerShell · Linux

EVERYDAY COMMANDS

Text in. Audio out.

Speak some text

Generate an MP3 in the outputs folder with the default voice.

uv run cli.py "Living the dream"

Read a document

Choose a voice and speed, export a WAV, then play it when finished.

uv run cli.py -f notes.txt -v bf_emma -s 1.1 --format wav --play

Name the output

Set a filename and suppress status messages for scripts.

uv run cli.py "Hello" --filename greeting --silent

DISCOVER & AUTOMATE

Useful controls.

Device selection defaults to the best available option: Metal on Mac, CUDA where supported, then CPU.

# Browse every installed voice
uv run cli.py --list-voices

# Inspect the selected accelerator
uv run cli.py --backend-info --device auto

# Emit machine-readable events
uv run cli.py "Hello" --json

# See every flag
uv run cli.py --help

On macOS or Linux, you can use the shorter bin/tts wrapper in place of uv run cli.py.

Need the full reference?

Flags, voices, development setup, and backend details live in the repository.

Open the docs