EURASIP Journal on Audio, Speech, and Music Processing

Table 11 Comparison of recognition errors by type between the first and second pass

From: Context-dependent factored language models

Error type	1st pass	2nd pass	Difference
Deletion	1446	1216	−230
Insertion	401	498	97
Substitution	4013	4043	30
Total	5860	5757	−103

Back to article page