Acoustic Model with Multiple Lexicon Types for Indonesian Speech Recognition

Figure 9

The %WER comparison for different GMM-HMM models using 30,000 unvalidated utterances.