In structural pattern recognition the median string has been established as a useful tool to represent a set of strings. However, its exact computation is complex and of high computational burden. In this paper we propose a new approach for the computation of median string based on string embedding. Strings are embedded into a vector space and the median is computed in the vector domain. We apply three different inverse transformations to go from the vector domain back to the string domain in order to obtain a final approximation of the median string. All of them are based on the weighted mean of a pair of strings. Experiments show that we succeed to compute good approximations of the median string. Peer Reviewed