The value of this features extracted from each window-level segment are compared to certain thresholds, which are also adapted to the SNR.

• 将从音频片段提取的特征参数值与门限值相比较进行语音／音乐的分类，其中的门限值与一定的信噪比相适应。