Acoustic Model with Multiple Lexicon Types for Indonesian Speech Recognition

Table 15

The performance of the models using 46,550 validated augmented vs. nonaugmented utterances.

Data type	GMM-HMM and TDNNF model (%WER)
	Monophone	Triphone DELTA + DELTA-DELTA	Triphone LDA + MLLT	Triphone SAT	TDNNF
Validated utterances	72.17	51.27	41.57	42.80	25.10
Validated augmented utterances	73.80	49.52	39.83	40.85	24.96