Skip to main content

Table 4 Performance of the proposed systems on the test set

From: Neural network-based non-intrusive speech quality assessment using attention pooling function

Stytems R RMSE
LSTM-max 0.961 0.283
LSTM-avgerage 0.953 0.365
LSTM-linear softmax 0.949 0.335
LSTM-attention 0.962 0.283
CNN-max 0.955 0.302
CNN-avgerage 0.959 0.370
CNN-linear softmax 0.961 0.317
CNN-attention 0.963 0.299
CNN-LSTM-max 0.964 0.273
CNN-LSTM-avgerage 0.965 0.283
CNN-LSTM-linear softmax 0.957 0.315
CNN-LSTM-attention 0.967 0.269