我已经实现了一些代码，我为10个不同的wav文件提取mfcc，每个文件都有一个数字的话语多次。
我设法创建了一个GMM，但我也想尝试SVM。测试是单独的wav文件。
我的问题是 - 我有10个不同的wav文件的MFCC功能。如何将它们提供给SVM分类器？
这是我用来提取每个wav文件的功能的代码：
def extract_feat(audio, rate):
mfcc_feat = mfcc(audio, rate)
#standardizing the features
mfcc_feat = preprocessing.scale(mfcc_feat)
#calculating the delta and double delta
delta_one = delta(mfcc_feat,2)
delta_two = delta(delta_one,2)
combined = np.hstack([mfcc_feat,delta_one, delta_two])
return combined
def createTrainingGMM(path):
#iterating through the training file and getting all wav files to train GMM
for filename in glob.glob(path):
#getting the sample rate value: 16000hz and the data read from wav file
sr_value, x_value = wav.read(filename)
#calling extract_feat which returns 39 mfcc features for each value in the vector
vector = extract_feat(x_value, sr_value)#default values
如何为SVM做类似的事情？