Acoustic Model with Multiple Lexicon Types for Indonesian Speech Recognition

Figure 10

The %WER comparison for different GMM-HMM models using 46,550 validated utterances.