A　smart　insole　based　on　pressure　sensing　arrays　is　a　simple　and　effective　means　of　gait　analysis　assist　in　the　assessment　of　human　movement　and　neurological　health.　However,　these　smart　insoles　usually　fail　to　combine　high　sensitivity　with　a　wide　detection　range,　making　them　only　suitable　for　people　within　a　certain　body　weight　range.　Here,　based　on　the　synergy　of　porous　and　air-gap　structures,　we　develop　a　high-performance　and　high　stability　smart　insole,　which　has　a　sensitivity　of　up　to　16.064　kPa(-1)　in　a　wide　pressure　range　of　0.170　Pa　to　248　kPa.　After　combined　with　Decision　Tree　machine　learning　model,　gait　classification　and　recognition　can　be　as　high　as　99.96%.　Based　on　these,　a　tap　dance　game　was　designed,　which　proves　its　ability　to　identify　individual　activities,　and　demonstrates　its　potential　of　application　in　the　field　of　human-computer　interaction　and　medical　engineering.