Using Morphological Data in Language Modeling for Serbian Large Vocabulary Speech Recognition

Table 2

Audio database overview.

Database part	Amount of data (h)	Amount of speech (h)	Male (sub) speakers	Female (sub) speakers
Audio books	168	140	208	190
Radio talk shows	179	150	350	70
Phone recordings	61	41	169	181
Total	408	331	727	441
For training	379	308	677	410