Tetrak makes archive collections readable and searchable, especially in scripts that mainstream software serves poorly.
Much of the world’s cultural history is looked after by small, independent libraries and community archives. They often hit two problems at once: off-the-shelf archive software can’t read or search their scripts, and the systems that can are priced for large institutions.
We work on both. We publish free, open-source tools that turn scans of old printed text into searchable text, starting with Armenian, and that make it possible to search across scripts, so someone who types a name in Latin letters finds the record written in Armenian, Georgian, Cyrillic or Syriac. And we work with collections teams directly: assessing what they hold, building and hosting search sites for their catalogues, and helping with the technical side of grant applications.
Free, open source, and yours to use without us.
Turns scans of old print into searchable text, on your own machine. It runs several OCR (optical character recognition) engines over each file, keeps the best reading, and sets aside anything it cannot trust for a person to check.
A recogniser we trained ourselves for printed Armenian, in both the Eastern and the Western language and either spelling. Free to install on its own, and the engine Tetrak OCR uses for Armenian pages.
Transliteration and cross-script search
Works out which script a piece of text is in, writes it in Latin letters by the library standards, and reduces every spelling of a name to one form so that a single search finds them all. Armenian, Georgian, Russian and Syriac.
Decides, before any image library or PDF renderer opens a file, whether it is one of the formats you meant to read and within the limits you expect, and says why not in a sentence. The security watch records which published flaws it refuses.
The tools are free and you can use them on your own. When you would rather have us do the work, we build and host search sites over catalogues, with searches that work across scripts and spellings; look at what a collection holds and what would make it findable; write the technical side of grant applications; and train collections teams on the tools. What we do for collections.
In 2026 the number of published vulnerabilities more than doubled, and in October Pillow fixed thirty-eight security flaws in one release. A scanning pipeline cannot stop using decoders, but it can decide, from a file's first bytes and its header, whether a decoder should see it at all. We built that decision for ourselves, then made it a library, and this is why
Eight libraries decode every scan we read, and in 2026 the flaws found in software like them more than doubled. Two people cannot read that stream by hand. This is how we turn it into a weekly issue, a bucket, a test and a published record, with a model doing the first reading, a person deciding, and nothing transcribed by hand
Ephrem the Syrian is catalogued as ʾAprém, Afrem, Ephraem and ܐܦܪܝܡ, and the script itself writes no vowels at all. We have added Syriac to tetrak-translit with a fold that keys on consonants alone, and this is what that buys, what it costs, and where we would welcome the help of people who read Syriac