Spoken keyword spotting via multi-lattice alignment

We propose a method for nding keywords in an audio database using a spoken query. Our method is based on per-forming a joint alignment between a phone lattice generated from a spoken utterance query and a second phone lattice repre-senting a long utterance needing to be searched. We implement this joint alignment procedure in a graphical models framework. We evaluate our system on TIMIT as well as on the Switch-board conversational telephone speech (CTS) corpus. Our re-sults show that a phone lattice representation of the spoken query achieves higher performance than using only the 1-best phone sequence representation