Convolutional Neural Networks to Facilitate the Continuous Recognition of Arabic Speech with Independent Speakers

Table 2

WER and CER results of CNN-LSTM using MFCC.

No. of MFCC features	WER		CER
	DS1 (%)	DS2 (%)	DS1 (%)	DS2 (%)
13 features	4.21	2.26	4.18	3.61
25 features	4.04	1.26	3.83	2.86
39 features	4.04	1.43	3.82	3.36

The bold values indicate the best results of CNN-LSTM using MFCC from WER and CER.