Introducing Vox Computatoria
A Private, Local Tool for Listening to Text and Articles
There’s something satisfying about having an article read aloud while you’re doing something else. Whether you’re commuting, cooking, or just prefer listening over reading, text-to-speech (TTS) can be incredibly useful. But most TTS tools come with trade-offs—either they send your text to the cloud, require a subscription, or are clunky to set up.
That’s why I built Vox Computatoria, a lightweight, private, and local tool that converts text and web articles into speech. Powered by Piper TTS, an open-source speech synthesis engine, Vox Computatoria runs entirely on your own machine—meaning your data stays private, and you don’t need an internet connection to generate high-quality speech.
At its core, it’s simple: enter some text or a URL, choose a voice, and get a downloadable audio file. The app extracts the article’s text, strips out unnecessary content, and generates speech using your selected voice. Since it runs locally, there’s no delay from sending data to a cloud service, and if you have an NVIDIA GPU, Piper can use CUDA acceleration to synthesize speech even faster.
I built this because I wanted a private, fast way to listen to articles without relying on commercial cloud-based APIs. Most existing tools either limit usage, track requests, or require internet access to function. Piper TTS provides great-sounding, open-source voices, and integrating it with Flask makes it easy to run from any device on my network.
Vox Computatoria is powered by Flask, served using Gunicorn, and managed via systemd for automatic startup. BeautifulSoup handles text extraction from web pages, while Piper TTS does the heavy lifting of speech synthesis. A virtual environment ensures dependencies are neatly contained, and deployment is as simple as cloning the repository, installing dependencies, downloading the voice models, and starting the server. If you want it running in the background, a systemd service makes it easy to start and stop.
One important note: if you’re using this to process articles from paywalled sites, keep it private. Since Vox Computatoria can extract full articles—including those behind login-based paywalls (if you have access)—making it public would effectively mean sharing your subscription with others. That’s not fair to you, and it’s certainly not fair to the publication. If you have a subscription, use it responsibly.
For now, Vox Computatoria does exactly what I need it to—convert text and articles into high-quality speech while running entirely on my own machine. If you want to try it out or contribute improvements, the code is available on GitHub.
