The authors investigate the number of speakers and the amount of data that is required for the development of useable speaker-independent speech-recognition systems in resource-scarce languages. Their experiments employ the Lwazi corpus, which contains speech in the eleven official languages of South Africa. They find that a surprisingly small number of speakers (fewer than 50) and around 10 to 20 hours of speech per language are sufficient for the purposes of acceptable phone-based recognition.
Reference:
Barnard, E, Davel, M and Van Heerden, C. ASR corpus design for resource-scarce languages. 10th Annual Conference of the International Speech Communication Association (Interspeech 2009), Brighton, UK, 6-10 September 2009, pp 2847-2850
Barnard, E., Davel, M., & Van Heerden, C. (2009). ASR corpus design for resource-scarce languages. ISCA. http://hdl.handle.net/10204/5576
Barnard, E, M Davel, and C Van Heerden. "ASR corpus design for resource-scarce languages." (2009): http://hdl.handle.net/10204/5576
Barnard E, Davel M, Van Heerden C, ASR corpus design for resource-scarce languages; ISCA; 2009. http://hdl.handle.net/10204/5576 .
