search

tzaiyang / SpeechEmoRec

Speech Emotion Recognition Using Deep Convolutional Neural Network and Discriminant Temporal Pyramid Matching
51 stars 18 forks source link

gen_segments_melspec 函数看不太懂 #6

Open sshzhang opened 5 years ago

sshzhang commented 5 years ago

你好,我有点看不懂gen_segments_melspec((X, window_size, overlap_sz))这个函数具体疑惑点如下

1)你能讲解一下window_size和overlap_sz的含义吗？我理解是窗口大小和重叠部分的大小. 2)append = np.zeros((64, (window_step - (X.shape[-1]-overlap_sz) % window_step))) 这个扩展容量我有点看不太懂，能麻烦你讲解一下吗?

sshzhang commented 5 years ago

在dtpm文件中 下面代码var_p和Niter没定义 print('solve_p') var_p = solve_p(learning_rate,Niter,var_p) print(var_p)

Louis-ZhangLe commented 5 years ago

你跑了程序吗 准确率怎么样啊 为什么我跑了只有30%左右

sshzhang commented 5 years ago

我git下来的项目, 报错不知道为什么