提出数据挖掘在清晰度客观评价中的应用。首先计算原始语音和失真语音在Bark域中的失真测度,然后利用数据挖掘方法建立了一个客观预测模型,完成参数失真测度到主观评分的映射,模型输出即为客观评分。仿真结果表明,使用该方法评价的客观评分与主观评分的相关度,当采用训练集样本时达到0.962以上,而采用测试样本时达到0.951以上,达到很好的预测效果。 Put forward the application of data mining in the objective evaluation of clarity. Firstly, the distortion measure of original speech and distorted speech in Bark domain is calculated. Then an objective prediction model is established by data mining method, and the mapping of parameter distortion measure to subjective score is completed. The model output is objective score. The simulation results show that the correlation between the objective score and the subjective score evaluated by this method reaches more than 0.962 when the training set samples are used, and reaches 0.951 when the test sample is adopted, thus achieving a very good predictive effect.