DLD: An Optimized Chinese Speech Recognition Model Based on Deep Learning

Table 3

Results of the network models on THCHS-30.

Acoustic model	Convolutional layer activation function	Fully connected layer activation function	WER (%)
DCNN	Sigmoid	ReLU	22.23
DCNN (ReLU)	ReLU	ReLU	21.76
DCNN-CTC	ReLU	ReLU	20.18