Abstract: A wearable translation device that provides real-time language translation without a network connection is provided. The wearable translation device picks up speech from a user in a first language using a microphone facing the user, translates it into a second language, and outputs synthesized speech in the second language through a speaker facing the listener. The use of large speakers allows for greater comprehensibility than with existing systems. In some embodiments, noise cancellation signals are output through a speaker facing the user to reduce the amount of the user's voice and ambient background noise that is audible to the listener. In some embodiments, the wearable translation device provides two-way translation.