The NCHLT speech corpus contains wide-band speech from approximately 200 speakers per language, in each of the eleven of cial languages of South Africa. We describe the design and development processes that were undertaken in order to develop the corpus, and report on associated materials such as orthographic transcriptions and pronunciation dictionaries that were released as part of the corpus. In order to benchmark speech recognition performance on the corpus, we have also developed both phone-recognition and word-recognition systems for all eleven languages; we nd that high accuracies can be achieved for these speaker-independent but vocabulary-dependent recognition tasks in all languages.
Reference:
Barnard, E, Davel, M.H, Van Heerden, C, De Wet, F and Badenhorst, J. 2014. The NCHLT speech corpus of the South African languages. In: 4th International Workshop on Spoken Language Technologies for Under-Resourced Languages, St Petersburg, Russia, 14-16 May 2014
Barnard, E., Davel, M., Van Heerden, C., De Wet, F., & Badenhorst, J. (2014). The NCHLT speech corpus of the South African languages. http://hdl.handle.net/10204/7549
Barnard, E, MH Davel, C Van Heerden, Febe De Wet, and J Badenhorst. "The NCHLT speech corpus of the South African languages." (2014): http://hdl.handle.net/10204/7549
Barnard E, Davel M, Van Heerden C, De Wet F, Badenhorst J, The NCHLT speech corpus of the South African languages; 2014. http://hdl.handle.net/10204/7549 .