ofxSupertonic

openFrameworks addon for fast, local text-to-speech using Supertonic and ONNX Runtime.

What is ofxSupertonic?

ofxSupertonic brings the Supertonic C++ ONNX text-to-speech pipeline into the openFrameworks ecosystem. It provides a C++ wrapper for loading the speech models and voice styles, synthesizing text, and writing the generated audio to WAV files.

This makes it possible to build low-latency, offline speech synthesis into creative coding projects and installations without relying on cloud services.

Tested Environments

ofxSupertonic has been validated on:

x86_64 (Ubuntu 24.04.3 LTS) ARM 64-bit (Raspberry Pi) macOS (Apple Silicon M2) Windows 11 openFrameworks: of_v0.12.1_linux64_gcc6_release

Setup

1. Install ONNX Runtime and Models

From the addon root, run:

python3 scripts/install.py

On Windows, use python instead if Python is installed under that command:

python scripts/install.py

The installer detects the current operating system and automatically:

1. Downloads and installs the appropriate ONNX Runtime libraries into libs/onnxruntime. 2. Downloads the Supertonic ONNX models and example voice styles into example_tts/bin/data/models.

The default model directory can be changed with --models-dir. The installer also supports --skip-libs, --skip-models, and --dry-run; run python3 scripts/install.py --help for details.

2. Build and Run example_tts

Ubuntu / Debian (Linux) and macOS:

From the addon root, build the example with:

cd example_tts
make

Run the release executable with:

make RunRelease

Windows:

Generate or update example_tts using the openFrameworks Project Generator. Open the generated Visual Studio solution (.sln), then build and run the example from Visual Studio.

License

Copyright (c) 2026 Yannick Hofmann.

BSD Simplified License.

For information on usage and redistribution, and for a DISCLAIMER OF ALL WARRANTIES, see the file LICENSE.txt in this distribution.