Research on English Vocabulary and Speech Corpus Recognition Based on Deep Learning

Figure 11

Comparison of recognition rates between MFCC features and DAE midlayer features (broken lines).