Skip to main content

Table 1 The corpus size and number of senones of the Babel languages

From: Advanced recurrent network-based hybrid acoustic models for low resource speech recognition

Language Training hours # senone Language Training hours # senone Language Training hours # senone
Cantonese(101) 140.7 4687 Assamese(102) 60.3 4707 Bengali(103) 61.1 4929
Pashto(104) 77.3 4823 Turkish(105) 76.4 4791 Tagalog(106) 83.8 4814
Vietnamese(107) 87.1 4692 Haitian(201) 66.5 4875 Swahili(202) 44.0 4638
Lao(203) 65.2 4667 Tamil(204) 64.1 4560 Kurdish (205) 41.7 4412
Zulu(206) 61.3 4464 Tok Pisin(207) 39.0 4565 Cebuano(301) 41.0 4603
Kazakh(302) 39.6 4714 Telugu(303) 41.8 4515 Lithuanian(304) 42.1 4755
Guarani(305) 42.6 4504 Igbo(306) 43.7 4659 Amharic(307) 43.2 4685
Mongolian(401) 45.7 4537 Javanese(402) 45.1 4763 Dholuo(403) 41.3 4571
  1. The number in the () indicates the language id per the Babel program
\