The accent was proved to be the essential links between linguistics and acoustics, and behaves as an important parameter for prosody processing and unit selection in speech synthesis system. In the paper, some acoustical measurements are carried out on F0, duration, silence in order to disclose the relationship between accent and corresponding acoustical parameters. The normalized acoustic parameters are induced to facilitate the accent detecting. Based on this, a model is proposed to predict accent in spontaneous speech. The method is proved to be very successful and has been used to label accent automatically for our large corpus. Further listening tests also show that the labelling results reaches 86% accurate rate, which is nearly the same as the agreement of hand labelling results. Furthermore, different native’s evaluations are also compared in the paper.