Abstract: The accuracy of speech recognition directly depends on accurate endpoint detection.Endpoint detec- tion is a very difficult task in the noise environment.It will be degraded with the decrease of SNR and differ- ent noise affects the accuracy of speech recognition.As a result,this paper proposed an endpoint detection ap- proach which is applicable to the telephone speech recognition system for city's name.The approach inte- grates band-partitioning spectral entropy and spectral energy to form a set of new feature parameters that can compensating the drawback of entropy and energy so that the performance of the detection is improved.