声调是汉语普通话中最重要的信息,随着汉语语音技术的进一步发展,声调的识别与评测已经成为汉语语音技术发展的一个主要方向。文章建立了一种基于GMM后验概率评估模型的汉语声调评估模型,并利用相关系数和分级错误率两个指标对模型进行实验分析。实验结果表明:这种模型很好的融合了声调基频特征和声调识别结果,具有一定的实用价值。 Tone is the most important message in Mandarin Chinese. With the further development of Chinese speech technology, tone recognition and evaluation have become a major direction in the development of Chinese speech technology. In this paper, a Chinese tone evaluation model based on the GMM posterior probability assessment model is established, and the model is experimentally analyzed by using the correlation coefficient and the classification error rate. The experimental results show that this model has a good combination of tone pitch features and tone recognition results, which has some practical value.