With any machine learning, the more data you have, the better the result,” says Ian Pattison, Google Cloud’s head of customer engineering in the UK and Ireland. “If you’re doing a translation from English to Arabic, there are millions of pre-existing translated phrases you can use to build your training set. ”įabricius draws and retouches the hieroglyphs, identifies them using Google Photos’ machine learning technology, and computes the sequencing of a particular phrase to come up with suggested translations. Those translations, however, aren’t currently done with machine learning because there’s insufficient source data. “Many hieroglyphs are visually very similar, so this is why this tool was made-to make it more efficient and faster because it currently takes a long time. “What does is take the idea of having a dictionary version of hieroglyphs and make a quicker way to identify based on visual similarity,” says Chance Coughenour, a digital archaeologist at Google Arts & Culture. Now, though, Google has decided to help our intrepid tomb raiders by creating “Fabricius,” a machine-learning-based online tool that can near-instantly identify and translate hieroglyphs from photographs, providing three potential translation matches in a matter of moments. “‘Happy birthday’ in Ancient Egyptian was phrased ‘glorious festival of your delivery’-while ‘hello’ translates to ‘greetings to you’.”

Also, full-stops are absent, so there’s no indication where a sentence begins or ends.

A big clue in reading them is that any people or animals will face the start of the sentence, but with hieroglyphs often incomplete or illegible, there’s no guarantee of such indicators being present. Those mischievous pyramid-builders amped up the confusion further by writing hieroglyphs left-to-right, right-to-left, and top-to-bottom depending on their mood.

Ancient glyph translator