CASIA汉语情感语料库由中国科学院自动化所（Institute of Automation, Chinese Academy of Sciences）录制，共包括四个专业发音人，六种情绪生气（angry）、高兴（happy）、害怕（fear）、悲伤（sad）、惊讶（surprise）和中性（neutral），共9600句不同发音。其中300句是相同文本的，也即是说对相同的文本赋以不同的情感来阅读，这些语料可以用来对比分析不同情感状态下的声学及韵律表现；另外100句是不同文本的，这些文本从字面意思就可以看出其情感归属，便于录音人更准确地表现出情感。我们选取300句相同文本的音频作为实验数据，取前200句为训练集，后100句为测试集。实验同样对数据集进行了切分处理。
CASIA Chinese emotional corpus was recorded by the Institute of Automation, Chinese Academy of Sciences. It includes four professional speakers and six kinds of emotions: angry, happy, sad, surprise and neutral, a total of 9600 different pronunciation. 300 of the corpus are the same text. That is to say the same text given different emotions to read, these corpus can be used to compare the analysis of different emotional state of the acoustic and rhythmic performance. The other 100 are different text, these text can be seen from the literal meaning of its emotional attribution, which make it easy to record people more accurately show the emotions. We selected the same text of the corpus as experimental data, and take the first 200 for the training set and the after 100 for the test set. The experiment also slices the data set.