Improvements in Tone Pronunciation Scoring for Strongly Accented Mandarin Speech

This paper discusses a tone pronunciation scoring system of Mandarin. It recognizes tones of syllables by using GMM model and uses the recognition results for tone assessment. Initially, experiment results are bad on strongly accented speech. There are two reasons: one is that the inaccurate force-alignment leads to incomplete F0 contours; the other is due to the special pattern of F0 contours. We propose several measures to the problems. The first is to make the extraction of F0 contour independent of the force-alignment. The second is to base the scoring on GMM posterior probabilities. The third is to use the same accented speech to train the GMM model. And the last is to train the fractionized bi-tone GMM models to cover tone changes in the multiplecharacter words. After these measures are taken, the tone scoring correct rate is improved from 60.2% to 83.3%.