Our first attempt at lattice generation is being implemented as a post-processing step for N-best hypotheses. The N-best list of hypotheses is cast into a lattice format, which can be used in the second pass of decoding to get the overall best hypothesis. In this talk we will present the motivation for lattice generation, review the basic steps for implementation of lattice generation, and review some recent work on the components of this process: grammar decoding and N-Best searching.