Tolle Lege — take it up and hear it
Turn any document into a spoken summary. On your own machine, with an AI you control, and nothing hosted by anyone else.
Honest about small amounts: a payment of €1 is almost entirely eaten by the fixed fees Gumroad and the card processor charge per transaction — roughly €0.90 of it, leaving small change for me. From about €3 up, most of what you pay actually arrives. Below that, take it for free instead. That is a real option, not a polite gesture.
The same job
$19.99
per month — about $240 a year, every year
Once, or nothing
no subscription, no account, no upsell
Why it is different
Run a local model through Ollama and nothing leaves your machine — no cloud, no quota, no tracking. Prefer a cloud model? Bring your own key; it lives in the Windows credential vault, never in our files.
Tell it five minutes and the summary is sized to fit under it — something the big players do not give you.
Summarize an English report into Dutch, French or German and the figures, names and technical terms stay right. It does not turn jargon into nonsense.
For a recipe or a manual, skip the summary entirely and have it read the document itself, word for word.
A natural English and a Flemish/Dutch voice ship with it and run locally. Other languages download on first use.
Listen in the app, or export the audio and take it with you on the commute.
How it works
Honest about your data
With a cloud provider, the text you summarize is sent to that provider to be summarized. Google’s free Gemini tier may use it to improve their products; a paid key does not.
With a local model, or with the Read aloud button, nothing leaves your machine at all. We do not run a server, collect data, or track anything.
Questions
Yes — pay what you want, including nothing. You bring your own AI key (free tiers exist) or run a model locally.
Windows 10 or 11, and an AI to do the summarizing: Google Gemini (free tier, no card needed), a local model via Ollama (unlimited, fully offline), or an OpenAI/Claude key you already have.
The voice always does. Summarizing needs an AI: a local model is fully offline, cloud models need internet.
.txt, .md, .docx and .pdf — or paste text directly. Scanned PDFs are images with no selectable text, so export those to text first.
About 307 MB. It bundles the speech engine, two voices and everything else it needs, so there is nothing further to install.
Free, local, privacy-first. No tracking, no ads, made in the EU.