音频分割出的每个音频段具有相同的声学特征,对每个音频段进取小波域MFC

Each audio segment from audio segmentation has the same acoustic characteristics, and the wavelet domain MFCC feature is used for each audio segment as the input of the clustering algorithm. However, most of these divided audio segments are not only one second, some even tens of seconds, how to extract the most important feature vector from these audio segments of different sizes to represent the entire audio segment data

Each audio segment divided out of audio has the same acoustic characteristics for each audio segment enterprising wavelet domain MFCC characteristics as input to the clustering algorithm. But most of these split audio segments, not just one second, some even tens of seconds, extract the most important feature vectors from these audio segments of varying sizes to represent the entire audio segment data

音频分割出的每个音频段具有相同的声学特征,对每个音频段进取小波域MFCC 特征,作为聚类算法的输入．但是这些被分割开的音频段,大多不只一秒,有些甚至是几十秒,如何从这些大小不一的音频段中提取出最重要的特征向量,来表示整个音频段数据<br>