Abstract
This paper proposes a continuous recognition algorithm for online interactive English speech segment based on artificial bee colony algorithm. Firstly, the speech signal is pre-weighted, divided into frames, added window and made endpoint detection, etc. The strong MFCC feature parameters are selected. The artificial bee colony algorithm is used to realize the continuous recognition of the speech segment. The results show that the algorithm is a good method for parameter optimization. It can not only overcome the local optimal solution, but also improve the recognition rate.