The speech evaluation method of English phoneme mobile learning system

An Android based English phoneme mobile learning platform is built in accordance with the characteristics of English speech. The speech evaluation method of feature comparison is created by using Melfrequency Cepstrum Coefficients and adopting dynamic time warping. An empirical study was made to test the effectiveness of the system on phoneme learning. 32 participants received phoneme training in the form of teacher's instruction and autonomous training through the software. The pre and post test showed their pronunciation improved significantly, Kendall's W shows that the system possesses high accuracy of grading for the phoneme and words pronunciation. Majority of the participants thought it was useful and convenient.