EURASIP Journal on Audio, Speech, and Music Processing

Table 5 Average real-time factors for recognition in the first pass obtained on the BNSI evaluation set

From: Context-dependent factored language models

Vocabulary size	Real-time factor
	Good-Turing		Knesser-Ney
	Bigram	Trigram	Bigram	Trigram
60K	6.29	18.46	6.14	18.62
100K	7.82	23.42	7.52	23.37
200K	10.44	31.55	10.48	31.71
300K	12.66	37.09	12.88	37.70

Back to article page