AlfaBetic, a startup that launched at TC 50, is offering a translation service that would allow publishers to get their content translated into ten of the world’s most widely spoken languages. According to the company, doing so would enable them to get their content across 842 million potential readers. That is certainly a huge number, but had they been able to translate the content just in Mandarin, it alone would have put our content in front of more then 1 billion eyeballs.
The service employs two phase parsing. First the text is translated via machines that is exactly what Translate.Google.com does. After an initial raw translation is accomplished the text is run through an actual human being who ensures that the translated text is free of any errors, omissions and commissions.
AlfaBetic would naturally be paying the human translators, but I guess they would be able to get cheap labor by using services like Amazon’s Mechanical Turk. The service also aims to reduce costs over time by making the automatic translations more accurate by employing “Statistical Machine Learning Techniques”, which would improve with each translated text.
If this was not all, AlfaBetic endeavors to preserve comments across the translated blog. So for example a comment posted on Startup Meme would be translated and posted to Startup Meme’s French based version and vice versa within a few minutes. Making TechCrunch a genie pig the service has translated it into 10 different languages already.
AlfaBetic wants to make money the old way by employing advertising. The company would be controlling advertising that would appear on the translated sites. The service promises to maximize revenues by bringing in local advertisers. Other than the advertising stuff, the service is completely free to use.
They are also looking into getting deals with local portals across the globe to help in syndicating the content to a wider audience. The company is based in Israel and has raised $250,000 in angel funding.
You can watch their TC 50 launch video below:
