Machine translation is a very tough problem and many have been doing a great job at improving its accuracy like Google, Babelfish and Sakhr. While we realize these tools solve a major problem, they don’t tackle one important piece: the flexibility to perfect the translated text.  Perhaps it’s because they assume that users translate text for the purpose of reading it rather then actually using it for a blog post, email, document etc …

A couple of weeks ago we asked you how often you use translation tools and how you use them. We listened to your answers and decided to hack a quick translation tool as an experiment in finding a more natural interaction.  We used Google’s translation API on the back end, while we focused on providing a free flowing end-user experience to allow them to perfect the translated text and easily move between the input and output.

This is still in the experimental stage so you might find a bunch of bugs but we would love to hear your feedback and find out whether you think we are moving in the right direction. Give it a try at  www.yamli.com/translate/