Acoustic Model with Multiple Lexicon Types for Indonesian Speech Recognition

Table 11

The performance of the acoustic models using 30,000 unvalidated utterances.

Lexicon type	GMM-HMM model (%WER)
	Monophone	Triphone DELTA + DELTA-DELTA	Triphone LDA + MLLT	Triphone SAT
1c-vocab-full	85.59	72.34	63.44	63.28
enmap-1c-vocab-full	85.34	72.35	64.33	64.46
enmap-kv-vocab-full	84.09	77.01	70.14	70.54
kv-vocab-full	82.15	77.36	70.56	71.00